From yj76@columbia.edu Wed, 28 Feb 2001 16:07:42 -0500
Date: Wed, 28 Feb 2001 16:07:42 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] compliering error
This is a multi-part message in MIME format.
------=_NextPart_000_001B_01C0A1A0.94609210
Content-Type: text/plain;
charset="gb2312"
Content-Transfer-Encoding: quoted-printable
It looks that ITK.dsw generated by CMake was changed, I got the below =
message after a update:
Is it relating to my MSVC++ setting or it is something due to CMake? It =
only happened today, used to be fine all the time.
--------------------Configuration: VXLNumerics - Win32 =
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib=20
LINK : fatal error LNK1181: cannot open input file =
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
VXLNumerics.lib - 1 error(s), 358 warning(s)
--------------------Configuration: ITKCommon - Win32 =
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib=20
LINK : fatal error LNK1181: cannot open input file =
"\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
Error executing link.exe.
ITKCommon.lib - 1 error(s), 1 warning(s)
--------------------Configuration: ITKBasicFilters - Win32 =
Debug--------------------
Compiling...
itkWriter.cxx
Creating library...
Microsoft (R) Library Manager Version 6.00.8447
Copyright (C) Microsoft Corp 1992-1998. All rights reserved.
LINK : fatal error LNK1181: cannot open input file =
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
ITKBasicFilters.lib - 1 error(s), 0 warning(s)
------=_NextPart_000_001B_01C0A1A0.94609210
Content-Type: text/html;
charset="gb2312"
Content-Transfer-Encoding: quoted-printable
It looks that ITK.dsw generated by =
CMake was=20
changed, I got the below message after a update:
Is it relating to my MSVC++ setting or =
it is=20
something due to CMake? It only happened today, used to be fine all the=20
time.
--------------------Configuration: =
VXLNumerics -=20
Win32 Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib =
LINK : fatal=20
error LNK1181: cannot open input file=20
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error =
executing=20
link.exe.
VXLNumerics.lib - 1 error(s), 358=20
warning(s)
--------------------Configuration: ITKCommon - Win32=20
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib=
LINK=20
: fatal error LNK1181: cannot open input file=20
"\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
Err=
or=20
executing link.exe.
ITKCommon.lib - 1 error(s), 1=20
warning(s)
--------------------Configuration: ITKBasicFilters - Win32 =
Debug--------------------
Compiling...
itkWriter.cxx
Creating =
library...
Microsoft=20
(R) Library Manager Version 6.00.8447
Copyright (C) Microsoft Corp =
1992-1998.=20
All rights reserved.
LINK : fatal error LNK1181: cannot open input =
file=20
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error =
executing=20
link.exe.
ITKBasicFilters.lib - 1 error(s), 0=20
warning(s)
------=_NextPart_000_001B_01C0A1A0.94609210--
From bill.hoffman@kitware.com Wed, 28 Feb 2001 16:32:40 -0500
Date: Wed, 28 Feb 2001 16:32:40 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] compliering error
--=====================_1220739290==_.ALT
Content-Type: text/plain; charset="us-ascii"
It is working for me here.
There have been some cmake changes recently.
Did you rebuild CMake itself and run the CMakeSetup program again?
-Bill
At 04:07 PM 2/28/2001 -0500, you wrote:
>It looks that ITK.dsw generated by CMake was changed, I got the below message after a update:
>Is it relating to my MSVC++ setting or it is something due to CMake? It only happened today, used to be fine all the time.
>
>
>--------------------Configuration: VXLNumerics - Win32 Debug--------------------
>Compiling...
>
>......
>
>\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib
>LINK : fatal error LNK1181: cannot open input file "\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
>Error executing link.exe.
>
>VXLNumerics.lib - 1 error(s), 358 warning(s)
>--------------------Configuration: ITKCommon - Win32 Debug--------------------
>Compiling...
>
>......
>
>\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib
>LINK : fatal error LNK1181: cannot open input file "\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
>Error executing link.exe.
>
>ITKCommon.lib - 1 error(s), 1 warning(s)
>--------------------Configuration: ITKBasicFilters - Win32 Debug--------------------
>Compiling...
>itkWriter.cxx
>Creating library...
>Microsoft (R) Library Manager Version 6.00.8447
>Copyright (C) Microsoft Corp 1992-1998. All rights reserved.
>LINK : fatal error LNK1181: cannot open input file "\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
>Error executing link.exe.
>
>ITKBasicFilters.lib - 1 error(s), 0 warning(s)
>
--=====================_1220739290==_.ALT
Content-Type: text/html; charset="us-ascii"
It is working for me here.
There have been some cmake changes recently.
Did you rebuild CMake itself and run the CMakeSetup program again?
-Bill
At 04:07 PM 2/28/2001 -0500, you wrote:
It looks
that ITK.dsw generated by CMake was changed, I got the below message
after a update:
Is it relating to my MSVC++ setting or it is
something due to CMake? It only happened today, used to be fine all the
time.
--------------------Configuration: VXLNumerics
- Win32 Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib
LINK : fatal error LNK1181: cannot open input file
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
VXLNumerics.lib - 1 error(s), 358
warning(s)
--------------------Configuration: ITKCommon - Win32
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib
LINK : fatal error LNK1181: cannot open input file
"\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
Error executing link.exe.
ITKCommon.lib - 1 error(s), 1 warning(s)
--------------------Configuration: ITKBasicFilters - Win32
Debug--------------------
Compiling...
itkWriter.cxx
Creating library...
Microsoft (R) Library Manager Version 6.00.8447
Copyright (C) Microsoft Corp 1992-1998. All rights reserved.
LINK : fatal error LNK1181: cannot open input file
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
ITKBasicFilters.lib - 1 error(s), 0
warning(s)
--=====================_1220739290==_.ALT--
From yj76@columbia.edu Wed, 28 Feb 2001 16:54:48 -0500
Date: Wed, 28 Feb 2001 16:54:48 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] compliering error
This is a multi-part message in MIME format.
------=_NextPart_000_0033_01C0A1A7.28B90B80
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
I did, but it still happened.
----- Original Message -----=20
From: Bill Hoffman=20
To: Yinpeng Jin=20
Cc: insight-Developers=20
Sent: Wednesday, February 28, 2001 4:32 PM
Subject: Re: [Insight-developers] compliering error
It is working for me here.
There have been some cmake changes recently. =20
Did you rebuild CMake itself and run the CMakeSetup program again?
-Bill
At 04:07 PM 2/28/2001 -0500, you wrote:
It looks that ITK.dsw generated by CMake was changed, I got the =
below message after a update:
Is it relating to my MSVC++ setting or it is something due to CMake? =
It only happened today, used to be fine all the time.
=20
=20
--------------------Configuration: VXLNumerics - Win32 =
Debug--------------------
Compiling...
=20
......
=20
\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib=20
LINK : fatal error LNK1181: cannot open input file =
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
=20
VXLNumerics.lib - 1 error(s), 358 warning(s)
--------------------Configuration: ITKCommon - Win32 =
Debug--------------------
Compiling...
=20
......
=20
\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib=20
LINK : fatal error LNK1181: cannot open input file =
"\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
Error executing link.exe.
=20
ITKCommon.lib - 1 error(s), 1 warning(s)
--------------------Configuration: ITKBasicFilters - Win32 =
Debug--------------------
Compiling...
itkWriter.cxx
Creating library...
Microsoft (R) Library Manager Version 6.00.8447
Copyright (C) Microsoft Corp 1992-1998. All rights reserved.
LINK : fatal error LNK1181: cannot open input file =
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
=20
ITKBasicFilters.lib - 1 error(s), 0 warning(s)
=20
------=_NextPart_000_0033_01C0A1A7.28B90B80
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
I did, but it still =
happened.
----- Original Message -----
From:=20
Bill=20
Hoffman
Cc: insight-Developers<=
/A>=20
Sent: Wednesday, February 28, =
2001 4:32=20
PM
Subject: Re: =
[Insight-developers]=20
compliering error
It is working for me here.
There have been some =
cmake=20
changes recently.
Did you rebuild CMake itself and run the=20
CMakeSetup program again?
-Bill
At 04:07 PM 2/28/2001 =
-0500, you=20
wrote:
It looks=20
that ITK.dsw generated by CMake was changed, I got the below message =
after a=20
update:
Is it relating to my =
MSVC++=20
setting or it is something due to CMake? It only happened today, =
used to be=20
fine all the time.
--------------------Configuration: VXLNumerics - Win32=20
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib =
LINK :=20
fatal error LNK1181: cannot open input file=20
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error=20
executing link.exe.
VXLNumerics.lib - 1 error(s), 358=20
warning(s)
--------------------Configuration: ITKCommon - Win32=20
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib=
=20
LINK : fatal error LNK1181: cannot open input file=20
=
"\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
Err=
or=20
executing link.exe.
ITKCommon.lib - 1 error(s), 1=20
warning(s)
--------------------Configuration: ITKBasicFilters - =
Win32=20
Debug--------------------
Compiling...
itkWriter.cxx
Creating library...
Microsoft (R) =
Library Manager=20
Version 6.00.8447
Copyright (C) Microsoft Corp 1992-1998. All =
rights=20
reserved.
LINK : fatal error LNK1181: cannot open input file=20
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error=20
executing link.exe.
ITKBasicFilters.lib - 1 error(s), 0=20
warning(s)
------=_NextPart_000_0033_01C0A1A7.28B90B80--
From lng@insightful.com Wed, 28 Feb 2001 14:34:27 -0800
Date: Wed, 28 Feb 2001 14:34:27 -0800
From: Lydia Ng lng@insightful.com
Subject: [Insight-developers] compliering error
I am having the same problem.
I have also done a clean build of CMake.
Lydia
----- Original Message -----
From: Yinpeng Jin
To: insight-developers@public.kitware.com
Sent: Wednesday, February 28, 2001 1:54 PM
Subject: Re: [Insight-developers] compliering error
I did, but it still happened.
----- Original Message -----
From: Bill Hoffman
To: Yinpeng Jin
Cc: insight-Developers
Sent: Wednesday, February 28, 2001 4:32 PM
Subject: Re: [Insight-developers] compliering error
It is working for me here.
There have been some cmake changes recently.
Did you rebuild CMake itself and run the CMakeSetup program again?
-Bill
At 04:07 PM 2/28/2001 -0500, you wrote:
It looks that ITK.dsw generated by CMake was changed, I got the below
message after a update:
Is it relating to my MSVC++ setting or it is something due to CMake? It only
happened today, used to be fine all the time.
--------------------Configuration: VXLNumerics - Win32
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib
LINK : fatal error LNK1181: cannot open input file
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
VXLNumerics.lib - 1 error(s), 358 warning(s)
--------------------Configuration: ITKCommon - Win32
Debug--------------------
Compiling...
......
\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib
LINK : fatal error LNK1181: cannot open input file
"\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
Error executing link.exe.
ITKCommon.lib - 1 error(s), 1 warning(s)
--------------------Configuration: ITKBasicFilters - Win32
Debug--------------------
Compiling...
itkWriter.cxx
Creating library...
Microsoft (R) Library Manager Version 6.00.8447
Copyright (C) Microsoft Corp 1992-1998. All rights reserved.
LINK : fatal error LNK1181: cannot open input file
"\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
Error executing link.exe.
ITKBasicFilters.lib - 1 error(s), 0 warning(s)
From yj76@columbia.edu Wed, 28 Feb 2001 17:40:57 -0500
Date: Wed, 28 Feb 2001 17:40:57 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] compliering error
This is what I did:
I found that itkcommon is depending on vxl_numerics, and vxl_numerics also
depends on itkcommon.
so I tried to remove the dependency of itkcommon on vxl_numerics, and
complier itkCommon
now I got some itkCommon.lib, and then I can complier all other libs.
I hope it works fine, I need to test.
----- Original Message -----
From: "Lydia Ng"
To: "Yinpeng Jin" ;
Sent: Wednesday, February 28, 2001 5:34 PM
Subject: Re: [Insight-developers] compliering error
> I am having the same problem.
> I have also done a clean build of CMake.
>
> Lydia
>
> ----- Original Message -----
> From: Yinpeng Jin
> To: insight-developers@public.kitware.com
> Sent: Wednesday, February 28, 2001 1:54 PM
> Subject: Re: [Insight-developers] compliering error
>
>
> I did, but it still happened.
> ----- Original Message -----
> From: Bill Hoffman
> To: Yinpeng Jin
> Cc: insight-Developers
> Sent: Wednesday, February 28, 2001 4:32 PM
> Subject: Re: [Insight-developers] compliering error
>
>
> It is working for me here.
>
> There have been some cmake changes recently.
> Did you rebuild CMake itself and run the CMakeSetup program again?
>
> -Bill
>
> At 04:07 PM 2/28/2001 -0500, you wrote:
>
> It looks that ITK.dsw generated by CMake was changed, I got the below
> message after a update:
> Is it relating to my MSVC++ setting or it is something due to CMake? It
only
> happened today, used to be fine all the time.
>
>
> --------------------Configuration: VXLNumerics - Win32
> Debug--------------------
> Compiling...
>
> ......
>
> \insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib
> LINK : fatal error LNK1181: cannot open input file
> "\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
> Error executing link.exe.
>
> VXLNumerics.lib - 1 error(s), 358 warning(s)
> --------------------Configuration: ITKCommon - Win32
> Debug--------------------
> Compiling...
>
> ......
>
> \insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib
> LINK : fatal error LNK1181: cannot open input file
> "\insightnew\InsightBuild\Code\Numerics\vxl\Debug\VXLNumerics.lib"
> Error executing link.exe.
>
> ITKCommon.lib - 1 error(s), 1 warning(s)
> --------------------Configuration: ITKBasicFilters - Win32
> Debug--------------------
> Compiling...
> itkWriter.cxx
> Creating library...
> Microsoft (R) Library Manager Version 6.00.8447
> Copyright (C) Microsoft Corp 1992-1998. All rights reserved.
> LINK : fatal error LNK1181: cannot open input file
> "\insightnew\InsightBuild\Code\Common\Debug\ITKCommon.lib"
> Error executing link.exe.
>
> ITKBasicFilters.lib - 1 error(s), 0 warning(s)
>
>
From bill.hoffman@kitware.com Wed, 28 Feb 2001 17:37:28 -0500
Date: Wed, 28 Feb 2001 17:37:28 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] compliering error
I have found the problem and am working on a fix.
I will post to the list once I fix it.
-Bill
At 02:34 PM 2/28/2001 -0800, Lydia Ng wrote:
>I am having the same problem.
>I have also done a clean build of CMake.
>
>Lydia
From bill.hoffman@kitware.com Wed, 28 Feb 2001 17:45:20 -0500
Date: Wed, 28 Feb 2001 17:45:20 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] compliering error
I have just checked in a fix.
You will have to rebuild CMake, and
update Insight. I had to change several CMakeLists.txt files.
Also, there was a problem in CMake that had static libraries
depending on each other.
-Bill
From ibanez@cs.unc.edu Wed, 28 Feb 2001 20:41:12 -0500 (EST)
Date: Wed, 28 Feb 2001 20:41:12 -0500 (EST)
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] mesh data structure problem.
Hi Ting,
my mistake, I should have said:
point.Value() = p;
for the iterator,
intead of :
point.Value( p )
In any case, that will only help you
with the speed of access.
I think that the real problem in you
code could be the allocation.
Did you add a line like:
GetPoints()->Reserve( 1000 );
before starting to get access to the
points ?
This line will allocate a number of
points (1000 in this case ) for you.
Hope that helps,
Luis
----------------------------------------
On Wed, 28 Feb 2001, Ting Chen wrote:
> but point.value take no parameter and just return the value stored at the
> location, any other suggestion?
>
> >
> >PointsContainerPointer myPoints = GetPoints();
> >PointsContainer::Iterator point = myPoints->Begin();
> >PointType p;
> >p = 0,0,0;
> >while( point != myPoints->End() )
> >{
> > point.Value( p );
> > ++point;
> >}
> >
> >
> >
> >Luis
> >
> >
> >-----------------------------------------------------------------------
> >
> >Ting Chen wrote:
> >
> >> I write the following code in my itkDeformableMesh class:
> >>
> >> ....
> >> float ds[3], d[3]={0, 0, 0};
> >> SetPoint(1, (PointType)d);
> >> GetPoint(1, (PointType*)ds);
> >> ....
> >>
> >> but I found out that ds={-1.074e+008, 0, 0} after the operation.
> >>
> >> if I do
> >> ....
> >> float ds[3], d[3]={0, 0, 0};
> >> SetPoint(0, (PointType)d);
> >> GetPoint(0, (PointType*)ds);
> >> ....
> >>
> >> ds = {-1.074e+008, -1.074e+008, -1.074e+008}
> >>
> >> it seems the first element in the container has problems here. what is
> the
> >> problem here?
> >> thanks
> >> ting
> >>
> >> _______________________________________________
> >> Insight-developers mailing list
> >> Insight-developers@public.kitware.com
> >> http://public.kitware.com/mailman/listinfo/insight-developers
> >
> >--
> >______________________________________________________________________
> >
> >Luis Ibanez
> >Research Assistant Professor - Division of Neurosurgery
> >University of North Carolina at Chapel Hill
> >CB# 7060, Chapel Hill, NC 27599
> >email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
> >phone : (919)-843-9961 fax : (919)-966-6627
> >______________________________________________________________________
>
>
--
-----------------------------------------------------------
Luis Ibanez
Research Assistant Professor - Division of NeuroSurgery
University of North Carolina at Chapel Hill
Chapel Hill, NC, 27599 http://www.cs.unc.edu/~ibanez
------------------------------------------------------------
From ibanez@cs.unc.edu Thu, 01 Mar 2001 10:36:19 -0500
Date: Thu, 01 Mar 2001 10:36:19 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkMesh GetClassName() and windows.h
Hi,
I've found the following problem on WinNT:
Using itkMesh in a larger application mixed
with other non-itk classes.
It happens that if you include :
#include
#include
in this order,
and declare a variable of type itkMesh,
the calls to GetPoints() generate the following error
concerning GetClassName() :
------------------- error message -----------------------------
:\usr\ibanez\src\insightalpha\insight\code\common\itkmesh.txx(78) :
error C2039: 'GetClassName' : is not a member of 'Mesh >'
h:\microsoft visual studio\vc98\include\xstring(583) : while
compiling class-template member function 'class itk::SmartPointer > >
__thiscall itk::Mesh >::GetPoints(void)'
h:\usr\ibanez\src\insightalpha\insight\code\common\itkmesh.txx(78) :
error C2039: 'GetClassName' : is not a member of 'Mesh >'
h:\microsoft visual studio\vc98\include\xstring(583) : while
compiling class-template member function 'class itk::SmartPointer > >
__thiscall itk::Mesh >::GetPoints(void)'
Error executing cl.exe.
------------------ end error message -------------------------------
But, if the order of includes are revesed, like:
#include
#include
The code compiles fine.
Looks like some definitions on windows.h
obscures the Macro declaration of itkTypeMacro.
I cannont avoid the inclusion of "windows.h" because
it is buried nested in includes from other package (FLTK)
from which I'm deriving classes.
Is there something we can do about it ?
Thanks
Luis
--
______________________________________________________________________
Luis Ibanez
Research Assistant Professor - Division of Neurosurgery
University of North Carolina at Chapel Hill
CB# 7060, Chapel Hill, NC 27599
email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
phone : (919)-843-9961 fax : (919)-966-6627
______________________________________________________________________
From bill.hoffman@kitware.com Thu, 01 Mar 2001 10:32:24 -0500
Date: Thu, 01 Mar 2001 10:32:24 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] itkMesh GetClassName() and windows.h
You should not include windows.h and mesh. I
have created a new file called itkWindows.h, if you include
that instead of windows.h it will get rid of the problem.
At 10:36 AM 3/1/2001 -0500, Luis Ibanez wrote:
>Hi,
>
>I've found the following problem on WinNT:
>
>Using itkMesh in a larger application mixed
>with other non-itk classes.
>
>It happens that if you include :
>
> #include
> #include
>
>in this order,
>
>and declare a variable of type itkMesh,
>the calls to GetPoints() generate the following error
>concerning GetClassName() :
>
>------------------- error message -----------------------------
>:\usr\ibanez\src\insightalpha\insight\code\common\itkmesh.txx(78) :
>error C2039: 'GetClassName' : is not a member of 'Meshitk::DefaultStaticMeshType >'
> h:\microsoft visual studio\vc98\include\xstring(583) : while
>compiling class-template member function 'class itk::SmartPointeritk::VectorContainer > >
>__thiscall itk::MeshultStaticMeshType >::GetPoints(void)'
>h:\usr\ibanez\src\insightalpha\insight\code\common\itkmesh.txx(78) :
>error C2039: 'GetClassName' : is not a member of 'Meshitk::DefaultStaticMeshType >'
> h:\microsoft visual studio\vc98\include\xstring(583) : while
>compiling class-template member function 'class itk::SmartPointeritk::VectorContainer > >
>__thiscall itk::MeshultStaticMeshType >::GetPoints(void)'
>Error executing cl.exe.
>
>------------------ end error message -------------------------------
>
>But, if the order of includes are revesed, like:
>
>#include
>#include
>
>The code compiles fine.
>
>Looks like some definitions on windows.h
>obscures the Macro declaration of itkTypeMacro.
>
>I cannont avoid the inclusion of "windows.h" because
>it is buried nested in includes from other package (FLTK)
>from which I'm deriving classes.
>
>
>
>Is there something we can do about it ?
>
>
>
>Thanks
>
>
>Luis
>
>
>
>
>
>
>
>--
>______________________________________________________________________
>
>Luis Ibanez
>Research Assistant Professor - Division of Neurosurgery
>University of North Carolina at Chapel Hill
>CB# 7060, Chapel Hill, NC 27599
>email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
>phone : (919)-843-9961 fax : (919)-966-6627
>______________________________________________________________________
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From ibanez@cs.unc.edu Thu, 01 Mar 2001 10:58:36 -0500
Date: Thu, 01 Mar 2001 10:58:36 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkMesh GetClassName() and windows.h
It works fine,
Thanks !
I can even keep the indirect include to "windows.h"
(for which I have no choice because it is included from fltk
headers) as long as itkWindows.h is included before
itkMesh.h, like:
#include
#include
#include
Is that because of the #undef GetClassName in itkWindows.h ?
Thanks,
Luis
Bill Hoffman wrote:
> You should not include windows.h and mesh. I
> have created a new file called itkWindows.h, if you include
> that instead of windows.h it will get rid of the problem.
From bill.hoffman@kitware.com Thu, 01 Mar 2001 11:59:13 -0500
Date: Thu, 01 Mar 2001 11:59:13 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] itkMesh GetClassName() and windows.h
Yes, Brad has just added that undef to itkWin32Header.h, so if you do
an update, you may be able to get rid of the include of itkWindows.h altogether.
-Bill
At 10:58 AM 3/1/2001 -0500, Luis Ibanez wrote:
>Is that because of the #undef GetClassName in itkWindows.h ?
>
>Thanks,
>
>Luis
From blezek@crd.ge.com Thu, 1 Mar 2001 14:58:01 -0500 (EST)
Date: Thu, 1 Mar 2001 14:58:01 -0500 (EST)
From: Daniel J. Blezek, Ph.D. blezek@crd.ge.com
Subject: [Insight-developers] Template parameters
Hi all,
I'm digging into the Insight code, and noticed some inconsistancies is
how template paramaters are being used. In one case:
template
class ITK_EXPORT NeighborhoodOperatorImageFilter :
public ImageToImageFilter< Image,
Image >
In another:
template
class ITK_EXPORT ShrinkImageFilter:
public ImageToImageFilter
In another:
template
class ITK_EXPORT ThresholdImageFilter:public ImageToImageFilter
Case:
1. Specified the pixel type and dimension
2. Specified both input and output image types, not really necessary,
shrink should not change the pixel type of the data
3. Use only one image type for both input and output
I think this raises a fundamental question about how to specify a filter.
Do we specify by pixel type and dimension, or by input/output type, or
just one type?
Ideas?
-dan
--
Daniel Blezek, Ph.D.
blezek@crd.ge.com
Visualization and Computer Vision Program
Electronic Systems Lab
GE Corporate Research & Development
From chenting@graphics.cis.upenn.edu Thu, 1 Mar 2001 16:01:36 -0600
Date: Thu, 1 Mar 2001 16:01:36 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] mesh data structure problem.
Hi! I have tried to add the memory allocation code before I do anything to
the pointcontainer, but I still cannot get the correct answer,
and by the way, did you try to add that piece of code into itkMeshTest and
run it?
ting
-----Original Message-----
From: Luis Ibanez
To: Ting Chen
Cc: Insight toolkit
Date: Thursday, March 01, 2001 8:17 AM
Subject: Re: [Insight-developers] mesh data structure problem.
>Hi Ting,
>
>my mistake, I should have said:
>
> point.Value() = p;
>
>for the iterator,
>intead of :
>
> point.Value( p )
>
>In any case, that will only help you
>with the speed of access.
>
>I think that the real problem in you
>code could be the allocation.
>Did you add a line like:
>
> GetPoints()->Reserve( 1000 );
>
>before starting to get access to the
>points ?
>
>This line will allocate a number of
>points (1000 in this case ) for you.
>
>
>Hope that helps,
>
>Luis
>
>
>----------------------------------------
>
>On Wed, 28 Feb 2001, Ting Chen wrote:
>
>> but point.value take no parameter and just return the value stored at the
>> location, any other suggestion?
>>
>> >
>> >PointsContainerPointer myPoints = GetPoints();
>> >PointsContainer::Iterator point = myPoints->Begin();
>> >PointType p;
>> >p = 0,0,0;
>> >while( point != myPoints->End() )
>> >{
>> > point.Value( p );
>> > ++point;
>> >}
>> >
>> >
>> >
>> >Luis
>> >
>> >
>> >-----------------------------------------------------------------------
>> >
>> >Ting Chen wrote:
>> >
>> >> I write the following code in my itkDeformableMesh class:
>> >>
>> >> ....
>> >> float ds[3], d[3]={0, 0, 0};
>> >> SetPoint(1, (PointType)d);
>> >> GetPoint(1, (PointType*)ds);
>> >> ....
>> >>
>> >> but I found out that ds={-1.074e+008, 0, 0} after the operation.
>> >>
>> >> if I do
>> >> ....
>> >> float ds[3], d[3]={0, 0, 0};
>> >> SetPoint(0, (PointType)d);
>> >> GetPoint(0, (PointType*)ds);
>> >> ....
>> >>
>> >> ds = {-1.074e+008, -1.074e+008, -1.074e+008}
>> >>
>> >> it seems the first element in the container has problems here. what is
>> the
>> >> problem here?
>> >> thanks
>> >> ting
>> >>
>> >> _______________________________________________
>> >> Insight-developers mailing list
>> >> Insight-developers@public.kitware.com
>> >> http://public.kitware.com/mailman/listinfo/insight-developers
>> >
>> >--
>> >______________________________________________________________________
>> >
>> >Luis Ibanez
>> >Research Assistant Professor - Division of Neurosurgery
>> >University of North Carolina at Chapel Hill
>> >CB# 7060, Chapel Hill, NC 27599
>> >email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
>> >phone : (919)-843-9961 fax : (919)-966-6627
>> >______________________________________________________________________
>>
>>
>
>--
>-----------------------------------------------------------
>Luis Ibanez
>Research Assistant Professor - Division of NeuroSurgery
>University of North Carolina at Chapel Hill
>Chapel Hill, NC, 27599 http://www.cs.unc.edu/~ibanez
>------------------------------------------------------------
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From brad.king@kitware.com Thu, 1 Mar 2001 17:36:52 -0500 (EST)
Date: Thu, 1 Mar 2001 17:36:52 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Template parameters
Here is my view:
> template
> class ITK_EXPORT NeighborhoodOperatorImageFilter :
> public ImageToImageFilter< Image,
> Image >
This version should never be used because it locks down the filter to
operating on the Image class directly. This prevents someone from using
an image adaptor or writing their own image class that conforms to our
image's interface.
> template
> class ITK_EXPORT ShrinkImageFilter:
> public ImageToImageFilter
This version should always be used. Any type can be passed to each
parameter, so image adaptors and other image classes can be substituted
for our image. Even when the image types must be the same for both input
and output, someone could still use an image adaptor on one end, and thus
we need two different template parameters.
> template
> class ITK_EXPORT ThresholdImageFilter:
> public ImageToImageFilter
This version should never be used due to the above argument about image
adaptors. If you want to make sure the data and dimensions are the same
for the two images, there are other ways to do it at compile time. I'll
try to come up with an example of this if someone needs to do it.
-Brad
From cates@cayenne.cs.utah.edu Thu, 1 Mar 2001 18:19:05 -0700 (MST)
Date: Thu, 1 Mar 2001 18:19:05 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] Template parameters
Seems like there was some discussion on this awhile back and people
agreed(?) that for most filters the was the
way to go. From a "black-box" component view of constructing a pipeline,
specifying input type and output type as the template parameters makes
good sense.
Downside is that it requires a lot of fancy typenaming of typedef'd
typedefs to get at the actual parameters you need. Might be wise to
establish some conventions for how to do this that are acceptable to all
the supported compilers. (ie. enum'ing the image dimension, getting the
pixel type, etc.)
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
On Thu, 1 Mar 2001, Daniel J. Blezek, Ph.D. wrote:
> Hi all,
>
> I'm digging into the Insight code, and noticed some inconsistancies is
> how template paramaters are being used. In one case:
>
> template
> class ITK_EXPORT NeighborhoodOperatorImageFilter :
> public ImageToImageFilter< Image,
> Image >
>
>
> In another:
>
> template
> class ITK_EXPORT ShrinkImageFilter:
> public ImageToImageFilter
>
>
> In another:
>
> template
> class ITK_EXPORT ThresholdImageFilter:public ImageToImageFilter
>
>
> Case:
> 1. Specified the pixel type and dimension
> 2. Specified both input and output image types, not really necessary,
> shrink should not change the pixel type of the data
> 3. Use only one image type for both input and output
>
> I think this raises a fundamental question about how to specify a filter.
> Do we specify by pixel type and dimension, or by input/output type, or
> just one type?
>
> Ideas?
> -dan
>
> --
> Daniel Blezek, Ph.D.
> blezek@crd.ge.com
> Visualization and Computer Vision Program
> Electronic Systems Lab
> GE Corporate Research & Development
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From lng@insightful.com Fri, 2 Mar 2001 00:09:06 -0800
Date: Fri, 2 Mar 2001 00:09:06 -0800
From: Lydia Ng lng@insightful.com
Subject: [Insight-developers] mesh data structure problem.
Hi,
The problem seems to be with VectorContainer::GetElementIfIndexExists()
it thinks id = 0 is invalid and fails ( see code below ).
Hence the bogus result when you call
mesh->GetPoint( 0, &point )
Can someone from GE/Kitware look into this?
Cheers,
Lydia
--------------- from itkVectorContainer.txx -------------------------
template
bool
VectorContainer< TElementIdentifier , TElement >
::GetElementIfIndexExists(ElementIdentifier id, Element* element) const
{
if((id > 0) && (id < this->VectorType::size()))
{
if(element)
{
*element = this->VectorType::operator[](id);
}
return true;
}
return false;
}
----- Original Message -----
From: "Ting Chen"
To: "Luis Ibanez"
Cc: "Insight toolkit"
Sent: Thursday, March 01, 2001 2:01 PM
Subject: Re: [Insight-developers] mesh data structure problem.
> Hi! I have tried to add the memory allocation code before I do anything to
> the pointcontainer, but I still cannot get the correct answer,
> and by the way, did you try to add that piece of code into itkMeshTest and
> run it?
> ting
> -----Original Message-----
> From: Luis Ibanez
> To: Ting Chen
> Cc: Insight toolkit
> Date: Thursday, March 01, 2001 8:17 AM
> Subject: Re: [Insight-developers] mesh data structure problem.
>
>
> >Hi Ting,
> >
> >my mistake, I should have said:
> >
> > point.Value() = p;
> >
> >for the iterator,
> >intead of :
> >
> > point.Value( p )
> >
> >In any case, that will only help you
> >with the speed of access.
> >
> >I think that the real problem in you
> >code could be the allocation.
> >Did you add a line like:
> >
> > GetPoints()->Reserve( 1000 );
> >
> >before starting to get access to the
> >points ?
> >
> >This line will allocate a number of
> >points (1000 in this case ) for you.
> >
> >
> >Hope that helps,
> >
> >Luis
> >
> >
> >----------------------------------------
> >
> >On Wed, 28 Feb 2001, Ting Chen wrote:
> >
> >> but point.value take no parameter and just return the value stored at
the
> >> location, any other suggestion?
> >>
> >> >
> >> >PointsContainerPointer myPoints = GetPoints();
> >> >PointsContainer::Iterator point = myPoints->Begin();
> >> >PointType p;
> >> >p = 0,0,0;
> >> >while( point != myPoints->End() )
> >> >{
> >> > point.Value( p );
> >> > ++point;
> >> >}
> >> >
> >> >
> >> >
> >> >Luis
> >> >
> >> >
> >>
>-----------------------------------------------------------------------
> >> >
> >> >Ting Chen wrote:
> >> >
> >> >> I write the following code in my itkDeformableMesh class:
> >> >>
> >> >> ....
> >> >> float ds[3], d[3]={0, 0, 0};
> >> >> SetPoint(1, (PointType)d);
> >> >> GetPoint(1, (PointType*)ds);
> >> >> ....
> >> >>
> >> >> but I found out that ds={-1.074e+008, 0, 0} after the operation.
> >> >>
> >> >> if I do
> >> >> ....
> >> >> float ds[3], d[3]={0, 0, 0};
> >> >> SetPoint(0, (PointType)d);
> >> >> GetPoint(0, (PointType*)ds);
> >> >> ....
> >> >>
> >> >> ds = {-1.074e+008, -1.074e+008, -1.074e+008}
> >> >>
> >> >> it seems the first element in the container has problems here. what
is
> >> the
> >> >> problem here?
> >> >> thanks
> >> >> ting
> >> >>
> >> >> _______________________________________________
> >> >> Insight-developers mailing list
> >> >> Insight-developers@public.kitware.com
> >> >> http://public.kitware.com/mailman/listinfo/insight-developers
> >> >
> >> >--
> >> >______________________________________________________________________
> >> >
> >> >Luis Ibanez
> >> >Research Assistant Professor - Division of Neurosurgery
> >> >University of North Carolina at Chapel Hill
> >> >CB# 7060, Chapel Hill, NC 27599
> >> >email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
> >> >phone : (919)-843-9961 fax : (919)-966-6627
> >> >______________________________________________________________________
> >>
> >>
> >
> >--
> >-----------------------------------------------------------
> >Luis Ibanez
> >Research Assistant Professor - Division of NeuroSurgery
> >University of North Carolina at Chapel Hill
> >Chapel Hill, NC, 27599 http://www.cs.unc.edu/~ibanez
> >------------------------------------------------------------
> >
> >
> >
> >_______________________________________________
> >Insight-developers mailing list
> >Insight-developers@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/insight-developers
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From lng@insightful.com Fri, 2 Mar 2001 00:31:20 -0800
Date: Fri, 2 Mar 2001 00:31:20 -0800
From: Lydia Ng lng@insightful.com
Subject: [Insight-developers] Visual C++ project tutorial
Hi Damion,
I found I also needed to set "Use Run-time Library" to
"Debug Multithreaded DLL" (debug mode ) and
"Multithreaded DLL" (release mode) else it won't compile.
In my system, these weren't the defaults for a console app.
Are you able compile using a different run-time library setting?
Cheers,
Lydia
I found that I also need to make
----- Original Message -----
From: "Damion Shelton"
To: "Insight-Developers"
Sent: Friday, February 23, 2001 1:21 PM
Subject: [Insight-developers] Visual C++ project tutorial
> Hi all...
>
> At Steve's request, I checked in a new tutorial file under
> Insight/Documentation called VC_Custom_Projects. It describes the project
> settings necessary to build a Visual C++ 6.0 project outside of the
Insight
> source tree. Please run through it if you have a chance and give me
feedback
> (particularly if there are errors!), as this sort of tutorial will be
useful
> to the participants in the alpha test. Thx.
>
> -Damion-
>
> /-----------------------------------------------\
> Damion Shelton
> University of Pittsburgh
> Department of Bioengineering
> dmsst59@pitt.edu
> 412-624-9931
> \-----------------------------------------------/
>
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From lorensen@crd.ge.com Fri, 2 Mar 2001 07:08:59 -0500
Date: Fri, 2 Mar 2001 07:08:59 -0500
From: Lorensen, William E (CRD) lorensen@crd.ge.com
Subject: [Insight-developers] mesh data structure problem.
I believe I introduced that problem when purging some signed/unsigned mis matches.
I checked in the fix.
Bill
-----Original Message-----
From: Lydia Ng [mailto:lng@insightful.com]
Sent: Friday, March 02, 2001 3:09 AM
To: Luis Ibanez
Cc: insight-developers@public.kitware.com
Subject: Re: [Insight-developers] mesh data structure problem.
Hi,
The problem seems to be with VectorContainer::GetElementIfIndexExists()
it thinks id = 0 is invalid and fails ( see code below ).
Hence the bogus result when you call
mesh->GetPoint( 0, &point )
Can someone from GE/Kitware look into this?
Cheers,
Lydia
--------------- from itkVectorContainer.txx -------------------------
template
bool
VectorContainer< TElementIdentifier , TElement >
::GetElementIfIndexExists(ElementIdentifier id, Element* element) const
{
if((id > 0) && (id < this->VectorType::size()))
{
if(element)
{
*element = this->VectorType::operator[](id);
}
return true;
}
return false;
}
----- Original Message -----
From: "Ting Chen"
To: "Luis Ibanez"
Cc: "Insight toolkit"
Sent: Thursday, March 01, 2001 2:01 PM
Subject: Re: [Insight-developers] mesh data structure problem.
> Hi! I have tried to add the memory allocation code before I do anything to
> the pointcontainer, but I still cannot get the correct answer,
> and by the way, did you try to add that piece of code into itkMeshTest and
> run it?
> ting
> -----Original Message-----
> From: Luis Ibanez
> To: Ting Chen
> Cc: Insight toolkit
> Date: Thursday, March 01, 2001 8:17 AM
> Subject: Re: [Insight-developers] mesh data structure problem.
>
>
> >Hi Ting,
> >
> >my mistake, I should have said:
> >
> > point.Value() = p;
> >
> >for the iterator,
> >intead of :
> >
> > point.Value( p )
> >
> >In any case, that will only help you
> >with the speed of access.
> >
> >I think that the real problem in you
> >code could be the allocation.
> >Did you add a line like:
> >
> > GetPoints()->Reserve( 1000 );
> >
> >before starting to get access to the
> >points ?
> >
> >This line will allocate a number of
> >points (1000 in this case ) for you.
> >
> >
> >Hope that helps,
> >
> >Luis
> >
> >
> >----------------------------------------
> >
> >On Wed, 28 Feb 2001, Ting Chen wrote:
> >
> >> but point.value take no parameter and just return the value stored at
the
> >> location, any other suggestion?
> >>
> >> >
> >> >PointsContainerPointer myPoints = GetPoints();
> >> >PointsContainer::Iterator point = myPoints->Begin();
> >> >PointType p;
> >> >p = 0,0,0;
> >> >while( point != myPoints->End() )
> >> >{
> >> > point.Value( p );
> >> > ++point;
> >> >}
> >> >
> >> >
> >> >
> >> >Luis
> >> >
> >> >
> >>
>-----------------------------------------------------------------------
> >> >
> >> >Ting Chen wrote:
> >> >
> >> >> I write the following code in my itkDeformableMesh class:
> >> >>
> >> >> ....
> >> >> float ds[3], d[3]={0, 0, 0};
> >> >> SetPoint(1, (PointType)d);
> >> >> GetPoint(1, (PointType*)ds);
> >> >> ....
> >> >>
> >> >> but I found out that ds={-1.074e+008, 0, 0} after the operation.
> >> >>
> >> >> if I do
> >> >> ....
> >> >> float ds[3], d[3]={0, 0, 0};
> >> >> SetPoint(0, (PointType)d);
> >> >> GetPoint(0, (PointType*)ds);
> >> >> ....
> >> >>
> >> >> ds = {-1.074e+008, -1.074e+008, -1.074e+008}
> >> >>
> >> >> it seems the first element in the container has problems here. what
is
> >> the
> >> >> problem here?
> >> >> thanks
> >> >> ting
> >> >>
> >> >> _______________________________________________
> >> >> Insight-developers mailing list
> >> >> Insight-developers@public.kitware.com
> >> >> http://public.kitware.com/mailman/listinfo/insight-developers
> >> >
> >> >--
> >> >______________________________________________________________________
> >> >
> >> >Luis Ibanez
> >> >Research Assistant Professor - Division of Neurosurgery
> >> >University of North Carolina at Chapel Hill
> >> >CB# 7060, Chapel Hill, NC 27599
> >> >email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
> >> >phone : (919)-843-9961 fax : (919)-966-6627
> >> >______________________________________________________________________
> >>
> >>
> >
> >--
> >-----------------------------------------------------------
> >Luis Ibanez
> >Research Assistant Professor - Division of NeuroSurgery
> >University of North Carolina at Chapel Hill
> >Chapel Hill, NC, 27599 http://www.cs.unc.edu/~ibanez
> >------------------------------------------------------------
> >
> >
> >
> >_______________________________________________
> >Insight-developers mailing list
> >Insight-developers@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/insight-developers
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
_______________________________________________
Insight-developers mailing list
Insight-developers@public.kitware.com
http://public.kitware.com/mailman/listinfo/insight-developers
From dmsst59@pitt.edu Fri, 2 Mar 2001 08:54:35 -0500
Date: Fri, 2 Mar 2001 08:54:35 -0500
From: Damion Shelton dmsst59@pitt.edu
Subject: [Insight-developers] visual c++ tutorial status
To Lydia and others....
It turns out there were two omissions in the tutorial (found by Stephen and
Bill) which prevent it from compiling if you follow it exactly. These are:
1) You need to use the multi-thread libraries, with /MDd for debug builds or
/MD for release builds
2) You need to enable RTTI, found on the "C++ language" pulldown under the
C/C++ tab in the project options window
Another suggestion provided by Stephen is adding the include and library
paths to Visual C++ in general, via Tools->Options->Directories, rather than
using the /I flags in the compiler settings for each project. In most
circumstances this is probably a good idea, because it saves you from having
to retype everything (probably not a good idea if you're going to be
distributing stand-alone example workspaces though).
Hopefully those two flags are all I overlooked. I'll be updating the
tutorial very soon, hopefully later today after I'm clear of my last
midterm. If anyone else has suggestions for additional things they'd like to
see, or any other errors/omissions, please let me know. Thanks for the help.
-Damion-
From cates@cayenne.cs.utah.edu Fri, 2 Mar 2001 15:22:06 -0700 (MST)
Date: Fri, 2 Mar 2001 15:22:06 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] hash_map
Hi all,
>From todays TCON it seems clear that we will need hash tables in Insight
and cannot use the non-standard STL hashed container extensions because
they are not supported by microsoft.
Our "itkHashTable" can either be completely STL-compatible, or
stand-alone, or somewhere in between. I think the latter is a good
compromise, ie. make a reasonable effort to implement the stl associative
container concept (ideally by modifying some public domain
implementation), but not try to tackle the full SGI hashed associative
container concept. Other folks thoughts on this?
If we agree that this is a reasonable way to go, I'll start putting
something together.
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
From chenting@graphics.cis.upenn.edu Fri, 2 Mar 2001 19:29:53 -0600
Date: Fri, 2 Mar 2001 19:29:53 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] mesh data structure problem.
Hi! Thanks for check in the fix, however, the problem has not been solved
yet.
when I use:
itk::Mesh Mesh;
Mesh::PointType x;
getpoint(i, x);
the value in x is still not right. please look into that.(though much better
now, all the data is there, but the order is not correct)
Thx
ting
I cannot get the proper value in the mesh
-----Original Message-----
From: Lorensen, William E (CRD)
To: 'Lydia Ng' ; Luis Ibanez
Cc: insight-developers@public.kitware.com
Date: Friday, March 02, 2001 6:09 AM
Subject: RE: [Insight-developers] mesh data structure problem.
>I believe I introduced that problem when purging some signed/unsigned mis
matches.
>
>I checked in the fix.
>
>Bill
>
>
>-----Original Message-----
>From: Lydia Ng [mailto:lng@insightful.com]
>Sent: Friday, March 02, 2001 3:09 AM
>To: Luis Ibanez
>Cc: insight-developers@public.kitware.com
>Subject: Re: [Insight-developers] mesh data structure problem.
>
>
>Hi,
>
>The problem seems to be with VectorContainer::GetElementIfIndexExists()
>it thinks id = 0 is invalid and fails ( see code below ).
>
>Hence the bogus result when you call
> mesh->GetPoint( 0, &point )
>
>Can someone from GE/Kitware look into this?
>
>Cheers,
>Lydia
>
>--------------- from itkVectorContainer.txx -------------------------
>template
>bool
>VectorContainer< TElementIdentifier , TElement >
>::GetElementIfIndexExists(ElementIdentifier id, Element* element) const
>{
> if((id > 0) && (id < this->VectorType::size()))
> {
> if(element)
> {
> *element = this->VectorType::operator[](id);
> }
> return true;
> }
> return false;
>}
>
>
>
>----- Original Message -----
>From: "Ting Chen"
>To: "Luis Ibanez"
>Cc: "Insight toolkit"
>Sent: Thursday, March 01, 2001 2:01 PM
>Subject: Re: [Insight-developers] mesh data structure problem.
>
>
>> Hi! I have tried to add the memory allocation code before I do anything
to
>> the pointcontainer, but I still cannot get the correct answer,
>> and by the way, did you try to add that piece of code into itkMeshTest
and
>> run it?
>> ting
>> -----Original Message-----
>> From: Luis Ibanez
>> To: Ting Chen
>> Cc: Insight toolkit
>> Date: Thursday, March 01, 2001 8:17 AM
>> Subject: Re: [Insight-developers] mesh data structure problem.
>>
>>
>> >Hi Ting,
>> >
>> >my mistake, I should have said:
>> >
>> > point.Value() = p;
>> >
>> >for the iterator,
>> >intead of :
>> >
>> > point.Value( p )
>> >
>> >In any case, that will only help you
>> >with the speed of access.
>> >
>> >I think that the real problem in you
>> >code could be the allocation.
>> >Did you add a line like:
>> >
>> > GetPoints()->Reserve( 1000 );
>> >
>> >before starting to get access to the
>> >points ?
>> >
>> >This line will allocate a number of
>> >points (1000 in this case ) for you.
>> >
>> >
>> >Hope that helps,
>> >
>> >Luis
>> >
>> >
>> >----------------------------------------
>> >
>> >On Wed, 28 Feb 2001, Ting Chen wrote:
>> >
>> >> but point.value take no parameter and just return the value stored at
>the
>> >> location, any other suggestion?
>> >>
>> >> >
>> >> >PointsContainerPointer myPoints = GetPoints();
>> >> >PointsContainer::Iterator point = myPoints->Begin();
>> >> >PointType p;
>> >> >p = 0,0,0;
>> >> >while( point != myPoints->End() )
>> >> >{
>> >> > point.Value( p );
>> >> > ++point;
>> >> >}
>> >> >
>> >> >
>> >> >
>> >> >Luis
>> >> >
>> >> >
>> >>
>>-----------------------------------------------------------------------
>> >> >
>> >> >Ting Chen wrote:
>> >> >
>> >> >> I write the following code in my itkDeformableMesh class:
>> >> >>
>> >> >> ....
>> >> >> float ds[3], d[3]={0, 0, 0};
>> >> >> SetPoint(1, (PointType)d);
>> >> >> GetPoint(1, (PointType*)ds);
>> >> >> ....
>> >> >>
>> >> >> but I found out that ds={-1.074e+008, 0, 0} after the operation.
>> >> >>
>> >> >> if I do
>> >> >> ....
>> >> >> float ds[3], d[3]={0, 0, 0};
>> >> >> SetPoint(0, (PointType)d);
>> >> >> GetPoint(0, (PointType*)ds);
>> >> >> ....
>> >> >>
>> >> >> ds = {-1.074e+008, -1.074e+008, -1.074e+008}
>> >> >>
>> >> >> it seems the first element in the container has problems here. what
>is
>> >> the
>> >> >> problem here?
>> >> >> thanks
>> >> >> ting
>> >> >>
>> >> >> _______________________________________________
>> >> >> Insight-developers mailing list
>> >> >> Insight-developers@public.kitware.com
>> >> >> http://public.kitware.com/mailman/listinfo/insight-developers
>> >> >
>> >> >--
>> >>
>______________________________________________________________________
>> >> >
>> >> >Luis Ibanez
>> >> >Research Assistant Professor - Division of Neurosurgery
>> >> >University of North Carolina at Chapel Hill
>> >> >CB# 7060, Chapel Hill, NC 27599
>> >> >email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
>> >> >phone : (919)-843-9961 fax : (919)-966-6627
>> >>
>______________________________________________________________________
>> >>
>> >>
>> >
>> >--
>> >-----------------------------------------------------------
>> >Luis Ibanez
>> >Research Assistant Professor - Division of NeuroSurgery
>> >University of North Carolina at Chapel Hill
>> >Chapel Hill, NC, 27599 http://www.cs.unc.edu/~ibanez
>> >------------------------------------------------------------
>> >
>> >
>> >
>> >_______________________________________________
>> >Insight-developers mailing list
>> >Insight-developers@public.kitware.com
>> >http://public.kitware.com/mailman/listinfo/insight-developers
>>
>>
>> _______________________________________________
>> Insight-developers mailing list
>> Insight-developers@public.kitware.com
>> http://public.kitware.com/mailman/listinfo/insight-developers
>>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From chenting@graphics.cis.upenn.edu Fri, 2 Mar 2001 19:53:03 -0600
Date: Fri, 2 Mar 2001 19:53:03 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] mesh data structure problem.
ok, I found out the problem is in type cast. now the getpoint method is ok!
Thanks Bill.
-----Original Message-----
From: Lorensen, William E (CRD)
To: 'Lydia Ng' ; Luis Ibanez
Cc: insight-developers@public.kitware.com
Date: Friday, March 02, 2001 6:09 AM
Subject: RE: [Insight-developers] mesh data structure problem.
>I believe I introduced that problem when purging some signed/unsigned mis
matches.
>
>I checked in the fix.
>
>Bill
>
>
>-----Original Message-----
>From: Lydia Ng [mailto:lng@insightful.com]
>Sent: Friday, March 02, 2001 3:09 AM
>To: Luis Ibanez
>Cc: insight-developers@public.kitware.com
>Subject: Re: [Insight-developers] mesh data structure problem.
>
>
>Hi,
>
>The problem seems to be with VectorContainer::GetElementIfIndexExists()
>it thinks id = 0 is invalid and fails ( see code below ).
>
>Hence the bogus result when you call
> mesh->GetPoint( 0, &point )
>
>Can someone from GE/Kitware look into this?
>
>Cheers,
>Lydia
>
>--------------- from itkVectorContainer.txx -------------------------
>template
>bool
>VectorContainer< TElementIdentifier , TElement >
>::GetElementIfIndexExists(ElementIdentifier id, Element* element) const
>{
> if((id > 0) && (id < this->VectorType::size()))
> {
> if(element)
> {
> *element = this->VectorType::operator[](id);
> }
> return true;
> }
> return false;
>}
>
>
>
>----- Original Message -----
>From: "Ting Chen"
>To: "Luis Ibanez"
>Cc: "Insight toolkit"
>Sent: Thursday, March 01, 2001 2:01 PM
>Subject: Re: [Insight-developers] mesh data structure problem.
>
>
>> Hi! I have tried to add the memory allocation code before I do anything
to
>> the pointcontainer, but I still cannot get the correct answer,
>> and by the way, did you try to add that piece of code into itkMeshTest
and
>> run it?
>> ting
>> -----Original Message-----
>> From: Luis Ibanez
>> To: Ting Chen
>> Cc: Insight toolkit
>> Date: Thursday, March 01, 2001 8:17 AM
>> Subject: Re: [Insight-developers] mesh data structure problem.
>>
>>
>> >Hi Ting,
>> >
>> >my mistake, I should have said:
>> >
>> > point.Value() = p;
>> >
>> >for the iterator,
>> >intead of :
>> >
>> > point.Value( p )
>> >
>> >In any case, that will only help you
>> >with the speed of access.
>> >
>> >I think that the real problem in you
>> >code could be the allocation.
>> >Did you add a line like:
>> >
>> > GetPoints()->Reserve( 1000 );
>> >
>> >before starting to get access to the
>> >points ?
>> >
>> >This line will allocate a number of
>> >points (1000 in this case ) for you.
>> >
>> >
>> >Hope that helps,
>> >
>> >Luis
>> >
>> >
>> >----------------------------------------
>> >
>> >On Wed, 28 Feb 2001, Ting Chen wrote:
>> >
>> >> but point.value take no parameter and just return the value stored at
>the
>> >> location, any other suggestion?
>> >>
>> >> >
>> >> >PointsContainerPointer myPoints = GetPoints();
>> >> >PointsContainer::Iterator point = myPoints->Begin();
>> >> >PointType p;
>> >> >p = 0,0,0;
>> >> >while( point != myPoints->End() )
>> >> >{
>> >> > point.Value( p );
>> >> > ++point;
>> >> >}
>> >> >
>> >> >
>> >> >
>> >> >Luis
>> >> >
>> >> >
>> >>
>>-----------------------------------------------------------------------
>> >> >
>> >> >Ting Chen wrote:
>> >> >
>> >> >> I write the following code in my itkDeformableMesh class:
>> >> >>
>> >> >> ....
>> >> >> float ds[3], d[3]={0, 0, 0};
>> >> >> SetPoint(1, (PointType)d);
>> >> >> GetPoint(1, (PointType*)ds);
>> >> >> ....
>> >> >>
>> >> >> but I found out that ds={-1.074e+008, 0, 0} after the operation.
>> >> >>
>> >> >> if I do
>> >> >> ....
>> >> >> float ds[3], d[3]={0, 0, 0};
>> >> >> SetPoint(0, (PointType)d);
>> >> >> GetPoint(0, (PointType*)ds);
>> >> >> ....
>> >> >>
>> >> >> ds = {-1.074e+008, -1.074e+008, -1.074e+008}
>> >> >>
>> >> >> it seems the first element in the container has problems here. what
>is
>> >> the
>> >> >> problem here?
>> >> >> thanks
>> >> >> ting
>> >> >>
>> >> >> _______________________________________________
>> >> >> Insight-developers mailing list
>> >> >> Insight-developers@public.kitware.com
>> >> >> http://public.kitware.com/mailman/listinfo/insight-developers
>> >> >
>> >> >--
>> >>
>______________________________________________________________________
>> >> >
>> >> >Luis Ibanez
>> >> >Research Assistant Professor - Division of Neurosurgery
>> >> >University of North Carolina at Chapel Hill
>> >> >CB# 7060, Chapel Hill, NC 27599
>> >> >email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
>> >> >phone : (919)-843-9961 fax : (919)-966-6627
>> >>
>______________________________________________________________________
>> >>
>> >>
>> >
>> >--
>> >-----------------------------------------------------------
>> >Luis Ibanez
>> >Research Assistant Professor - Division of NeuroSurgery
>> >University of North Carolina at Chapel Hill
>> >Chapel Hill, NC, 27599 http://www.cs.unc.edu/~ibanez
>> >------------------------------------------------------------
>> >
>> >
>> >
>> >_______________________________________________
>> >Insight-developers mailing list
>> >Insight-developers@public.kitware.com
>> >http://public.kitware.com/mailman/listinfo/insight-developers
>>
>>
>> _______________________________________________
>> Insight-developers mailing list
>> Insight-developers@public.kitware.com
>> http://public.kitware.com/mailman/listinfo/insight-developers
>>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From wlorens1@nycap.rr.com Sun, 04 Mar 2001 08:21:15 -0500
Date: Sun, 04 Mar 2001 08:21:15 -0500
From: Bill Lorensen wlorens1@nycap.rr.com
Subject: [Insight-developers] Your new filter
Julien,
Good to see your new filter that computes a Danielsson distance map. I'm assuming that you built and tested it on Windows. The other platforms are a bit more strict in some C++ usage. Most of the compile errors on the SGI and gnu compilers will go away if you use the typename hint. For example:
TOutputImage::PixelType replaced with typename TOutputImage::PixelType
Also, the sgi is confused with your usage of sqrt. The arguments are of type short. Is this really what you meant? The squaring of the short's could cause overflow.
Also, neither your .h nor .txx files ended with a new line.
Lastly, I believe the filter should be called:
itkDanielssonImageFilter
or perhaps
itkDanielssonDistanceImageFilter
according to the new naming conventions we discussed in Salt Lake.
Thanks,
Bill
From bahrahm@yahoo.com Mon, 5 Mar 2001 08:47:45 -0800 (PST)
Date: Mon, 5 Mar 2001 08:47:45 -0800 (PST)
From: Jisung Kim bahrahm@yahoo.com
Subject: [Insight-developers] New PhysicalImage[Adaptor] class
Hello.
I'm Jisung Kim and working at CADD Lab at UNC-CH.
I just made some changes to introduce the new
PhysicalImage class and
PhysicalImageAdaptor class.
SUMMARY
1. moved all space and origin related functions,
member variables,
and typedefs from ImageBase, Image, &
ImageAdaptor classes to
PhysicalImage & PhysicalImageAdaptor classes.
2. added PhysicalImage & PhysicalImageAdaptor class
files to
Insight/Code/Common directory.
3. changed some test programs (see CHANGES DETAIL):
4. deleted all space and origin related functions
from
itkImportImageFilter (in
Insight/Code/BasicFilters/) class, and
created itkImportPhysicalImageFilter with the
space and origin
related functions. (I just read through the
original class and
deleted those functions. So you might get some
run-time error. IF
YOU KNOW POSSIBLE RUNTIME ERRORS WITH THIS
MODIFICATIONS, please
let me know.)
5. changed the internal image type definition in the
ImageMomentsCalculator class from Image to
PhysicalImage. (It
seems that the origin and spacing are integral
parts of the
class. So, it would be nice to change the name of
the class to
PhysicalImageMomentsCalculator. I will ask the
person who is in
charge of maintaing CVS repository to rename the
class and make
appropriate changes.)
POSSIBLE PROBLEMS & SUGGESTIONS
1. Compile error in your own code (error messages
complain about
[Set|Get][Origin|Spacing] functions)
-> It might be caused by using
[Set|Get][Origin|Spacing]()
functions of Image class. Use PhysicalImage class
instead of Image
class. same for ImageAdaptor class.
2. Compile time error in Insight toolkit (error
messages complain about
[Set|Get][Origin|Spacing] functions)
->It might be caused by using ImportImageFilter
with Origin or
Spacing stuff. Use ImportPhysicalImageFilter or
class with
PhysicalImage class.
3. Runtime error with ImportImageFilter class
->As I mentioned in SUMMARY #4, it might be
causes by the
modification in both classes. please email me.
CHANGES DETAIL
* replaced Image class with PhysicalImage class
o
Insight/Testing/Code/Common/itkAffineTransform.cxx
o
Insight/Testing/Code/Common/itkImageIteratorTest.cxx
o Insight/Testing/Code/Common/itkIteratorTests.cxx
o
Insight/Testing/Code/Common/itkPixelAcessTest.cxx
o
Insight/Testing/Code/Common/itkBasicArchitectureTest.cxx
o
Insight/Testing/Code/BasicFilters/itkCyclicReference.cxx
o
Insight/Testing/Code/BasicFilters/itkShrinkImageTest.cxx
o
Insight/Testing/Code/BasicFilters/itkThresholdImageFilterTest.cxx
o
Insight/Testing/Code/Algorithms/itkCurvatureFlowTest.cxx
o
Insight/Testing/Code/Algorithms/itkInterpolateTest.cxx
o
Insight/Testing/Code/Algorithms/itkPrincipalAxesResampler.cxx
o
Insight/Testing/Code/Algorithms/itkImageMomentsTest.cxx
* also replaced ImageAdaptor class with
PhysicalImageAdaptor class
o
Insight/Testing/Code/BasicFilters/itkImageAdaptorPipelineTest.cxx
* special cases (see summary #4)
o
Insight/Testing/Code/BasicFilters/itkImportImageTest.cxx
- replaced Image class with PhysicalImage and
ImportImageFilter
Thank you,
=====
Jisung Kim
bahrahm@yahoo.com
106 Mason Farm Rd.
129 Radiology Research Lab., CB# 7515
Univ. of North Carolina at Chapel Hill
Chapel Hill, NC 27599-7515
__________________________________________________
Do You Yahoo!?
Get email at your own domain with Yahoo! Mail.
http://personal.mail.yahoo.com/
From cates@cayenne.cs.utah.edu Tue, 6 Mar 2001 13:07:49 -0700 (MST)
Date: Tue, 6 Mar 2001 13:07:49 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] changes to neighborhood base classes
Hello,
In the spirit of itk::Image, I've templated the itk::Neighborhood class
over allocator type. Didn't use stl allocators but rather modeled a
default allocator after vnl_vector. This means you can plug vnl_vectors
into neighborhood objects (iterators, neighborhoods, neighborhood
operators). Also neighborhoods now define const iterators through their
data.
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
From lorensen@crd.ge.com Wed, 7 Mar 2001 11:21:01 -0500
Date: Wed, 7 Mar 2001 11:21:01 -0500
From: Lorensen, William E (CRD) lorensen@crd.ge.com
Subject: [Insight-developers] changes to neighborhood base classes
Josh,
I checked in a few changes for SGI compatibility.
Also, the test itkGradientAnisoptropicDiffusionTest is in BasicFilters, yet the class is in
Algorithms.
I removed the test from the CMakeLists.txt file for now until it is resolved. Should the class be
moved to BasicFilters?
Bill
> -----Original Message-----
> From: Joshua Cates [SMTP:cates@cayenne.cs.utah.edu]
> Sent: Tuesday, March 06, 2001 3:08 PM
> To: Insight-Developers
> Subject: [Insight-developers] changes to neighborhood base classes
>
> Hello,
>
> In the spirit of itk::Image, I've templated the itk::Neighborhood class
> over allocator type. Didn't use stl allocators but rather modeled a
> default allocator after vnl_vector. This means you can plug vnl_vectors
> into neighborhood objects (iterators, neighborhoods, neighborhood
> operators). Also neighborhoods now define const iterators through their
> data.
>
>
> Josh.
>
> ______________________________
> Josh Cates
> School of Computer Science
> University of Utah
> Email: cates@cs.utah.edu
> Phone: (801) 587-7697
> URL: www.cs.utk.edu/~cates
>
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
From cates@cayenne.cs.utah.edu Wed, 7 Mar 2001 10:46:39 -0700 (MST)
Date: Wed, 7 Mar 2001 10:46:39 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] changes to neighborhood base classes
Hi Bill,
Thanks. I'm working on cleaning up some of the new warnings now.
Yes, the itkGradientAnisotropicDiffusionImageFilter.h/.txx should be in
BasicFilters. I'll go ahead and move it.
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
On Wed, 7 Mar 2001, Lorensen, William E (CRD) wrote:
> Josh,
> I checked in a few changes for SGI compatibility.
>
> Also, the test itkGradientAnisoptropicDiffusionTest is in BasicFilters, yet the class is in
> Algorithms.
> I removed the test from the CMakeLists.txt file for now until it is resolved. Should the class be
> moved to BasicFilters?
>
> Bill
>
>
> > -----Original Message-----
> > From: Joshua Cates [SMTP:cates@cayenne.cs.utah.edu]
> > Sent: Tuesday, March 06, 2001 3:08 PM
> > To: Insight-Developers
> > Subject: [Insight-developers] changes to neighborhood base classes
> >
> > Hello,
> >
> > In the spirit of itk::Image, I've templated the itk::Neighborhood class
> > over allocator type. Didn't use stl allocators but rather modeled a
> > default allocator after vnl_vector. This means you can plug vnl_vectors
> > into neighborhood objects (iterators, neighborhoods, neighborhood
> > operators). Also neighborhoods now define const iterators through their
> > data.
> >
> >
> > Josh.
> >
> > ______________________________
> > Josh Cates
> > School of Computer Science
> > University of Utah
> > Email: cates@cs.utah.edu
> > Phone: (801) 587-7697
> > URL: www.cs.utk.edu/~cates
> >
> >
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From brad.king@kitware.com Wed, 7 Mar 2001 14:37:42 -0500 (EST)
Date: Wed, 7 Mar 2001 14:37:42 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] itkTypeMacro
Hello all,
I have noticed warnings about classes with virtual functions but no
virtual destructor. The problem comes from the use of itkTypeMacro in
these classes. The macro is designed for use only in classes that are
part of the itk::LightObject hierarchy. Classes like itk::Vector should
not use the macro because they are not subclasses of the light object.
If anyone has written any classes with this problem, please remove the
macro from them. This will clear up several warnings from the dashboard.
Thanks,
-Brad
From dmsst59@pitt.edu Wed, 7 Mar 2001 17:00:02 -0500
Date: Wed, 7 Mar 2001 17:00:02 -0500
From: Damion Shelton dmsst59@pitt.edu
Subject: [Insight-developers] VC++ tutorial update
The VC++ tutorial has been updated to reflect last week's revelations
regarding multithreaded DLLs and RTTI. Let me know if any additional info
needs to be added.
-Damion-
From ibanez@cs.unc.edu Wed, 07 Mar 2001 20:58:42 -0500
Date: Wed, 07 Mar 2001 20:58:42 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkTypeMacro
Hi
The itkTypeMacro was removed also from itkCovariantVector
Thanks
Luis
--------------------------
Brad King wrote:
> Hello all,
>
> I have noticed warnings about classes with virtual functions but no
> virtual destructor. The problem comes from the use of itkTypeMacro in
> these classes. The macro is designed for use only in classes that are
> part of the itk::LightObject hierarchy. Classes like itk::Vector should
> not use the macro because they are not subclasses of the light object.
> If anyone has written any classes with this problem, please remove the
> macro from them. This will clear up several warnings from the dashboard.
>
> Thanks,
> -Brad
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
From ibanez@cs.unc.edu Wed, 07 Mar 2001 22:34:28 -0500
Date: Wed, 07 Mar 2001 22:34:28 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkIndex signed -> unsigned
Hi,
Recently itkIndex changed from using signed to unsigned values.
That has the advantage to eliminate a lot of warnings when it is
compared against image size (in most iterator loops),
but has the disadvantage that negative positions can't be specified.
we found that for the Danielsson distance map filter a signed index
was quite useful.
Should we create another class to express vector distances
between indices ?
(the equivalent of what the vector class is for the point class).
Thanks
Luis
From millerjv@crd.ge.com Thu, 8 Mar 2001 08:30:47 -0500
Date: Thu, 8 Mar 2001 08:30:47 -0500
From: Miller, James V (CRD) millerjv@crd.ge.com
Subject: [Insight-developers] itkIndex signed -> unsigned
I think we had discussed having a class "itkOffset" that could represent the difference between two
itkIndex's.
Anyone else have a suggestion for a name?
-----Original Message-----
From: Luis Ibanez [mailto:ibanez@cs.unc.edu]
Sent: Wednesday, March 07, 2001 10:34 PM
To: insight-developers@public.kitware.com
Subject: [Insight-developers] itkIndex signed -> unsigned
Hi,
Recently itkIndex changed from using signed to unsigned values.
That has the advantage to eliminate a lot of warnings when it is
compared against image size (in most iterator loops),
but has the disadvantage that negative positions can't be specified.
we found that for the Danielsson distance map filter a signed index
was quite useful.
Should we create another class to express vector distances
between indices ?
(the equivalent of what the vector class is for the point class).
Thanks
Luis
_______________________________________________
Insight-developers mailing list
Insight-developers@public.kitware.com
http://public.kitware.com/mailman/listinfo/insight-developers
From aylward@unc.edu Thu, 08 Mar 2001 12:12:26 -0500
Date: Thu, 08 Mar 2001 12:12:26 -0500
From: Stephen R. Aylward aylward@unc.edu
Subject: [Insight-developers] itkIndex signed -> unsigned
Hi,
I thought we had decided it would be best to keep itkIndex signed so
that we could check for <0 for termination of certain methods and so
that we can check offsets via subtraction and <0.
Stephen
Luis Ibanez wrote:
>
> Hi,
>
> Recently itkIndex changed from using signed to unsigned values.
> That has the advantage to eliminate a lot of warnings when it is
> compared against image size (in most iterator loops),
> but has the disadvantage that negative positions can't be specified.
> we found that for the Danielsson distance map filter a signed index
> was quite useful.
>
> Should we create another class to express vector distances
> between indices ?
> (the equivalent of what the vector class is for the point class).
>
> Thanks
>
> Luis
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
--
===============================================
Stephen R. Aylward
Assistant Professor of Radiology
Adjunct Assistant Professor of Computer Science
http://www.cs.unc.edu/~aylward
aylward@unc.edu
(919) 966-9695
From cates@cayenne.cs.utah.edu Thu, 8 Mar 2001 11:48:07 -0700 (MST)
Date: Thu, 8 Mar 2001 11:48:07 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] slice.h
Hello,
Where does slice.h reside on windows machines with VC++? I need a
cross-platform way to #include "std/slice.h" (see todays dashboard under
Win-NT-4.0-VC++)
Thanks,
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
From cates@cayenne.cs.utah.edu Thu, 8 Mar 2001 12:04:43 -0700 (MST)
Date: Thu, 8 Mar 2001 12:04:43 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] itkIndex signed -> unsigned
Hi,
I agree that we should keep itkIndex signed, for the reasons Stephen has
mentioned.
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
On Thu, 8 Mar 2001, Stephen R. Aylward wrote:
> Hi,
>
> I thought we had decided it would be best to keep itkIndex signed so
> that we could check for <0 for termination of certain methods and so
> that we can check offsets via subtraction and <0.
>
> Stephen
>
> Luis Ibanez wrote:
> >
> > Hi,
> >
> > Recently itkIndex changed from using signed to unsigned values.
> > That has the advantage to eliminate a lot of warnings when it is
> > compared against image size (in most iterator loops),
> > but has the disadvantage that negative positions can't be specified.
> > we found that for the Danielsson distance map filter a signed index
> > was quite useful.
> >
> > Should we create another class to express vector distances
> > between indices ?
> > (the equivalent of what the vector class is for the point class).
> >
> > Thanks
> >
> > Luis
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
>
> --
> ===============================================
> Stephen R. Aylward
> Assistant Professor of Radiology
> Adjunct Assistant Professor of Computer Science
> http://www.cs.unc.edu/~aylward
> aylward@unc.edu
> (919) 966-9695
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From bill.hoffman@kitware.com Thu, 08 Mar 2001 13:29:54 -0500
Date: Thu, 08 Mar 2001 13:29:54 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] slice.h
What is slice.h???
At 11:48 AM 3/8/2001 -0700, Joshua Cates wrote:
>Hello,
>
>Where does slice.h reside on windows machines with VC++? I need a
>cross-platform way to #include "std/slice.h" (see todays dashboard under
>Win-NT-4.0-VC++)
>
>Thanks,
>
>Josh.
>
>______________________________
> Josh Cates
> School of Computer Science
> University of Utah
> Email: cates@cs.utah.edu
> Phone: (801) 587-7697
> URL: www.cs.utk.edu/~cates
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From brad.king@kitware.com Thu, 8 Mar 2001 13:47:27 -0500 (EST)
Date: Thu, 8 Mar 2001 13:47:27 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] slice.h
According to the C++ standard, the "slice" class is defined by
#include
I don't know for sure if this works on all platforms, though.
-Brad
At 11:48 AM 3/8/2001 -0700, Joshua Cates wrote:
>Hello,
>
>Where does slice.h reside on windows machines with VC++? I need a
>cross-platform way to #include "std/slice.h" (see todays dashboard under
>Win-NT-4.0-VC++)
>
>Thanks,
>
>Josh.
From cates@cayenne.cs.utah.edu Thu, 8 Mar 2001 13:30:54 -0700 (MST)
Date: Thu, 8 Mar 2001 13:30:54 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] slice.h
Hi,
Sounds like I should just #include then to conform to the
standard. This has worked to define std::slice on VC++ in the past.
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
On Thu, 8 Mar 2001, Brad King wrote:
> According to the C++ standard, the "slice" class is defined by
> #include
>
> I don't know for sure if this works on all platforms, though.
> -Brad
>
> At 11:48 AM 3/8/2001 -0700, Joshua Cates wrote:
> >Hello,
> >
> >Where does slice.h reside on windows machines with VC++? I need a
> >cross-platform way to #include "std/slice.h" (see todays dashboard under
> >Win-NT-4.0-VC++)
> >
> >Thanks,
> >
> >Josh.
>
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From wlorens1@nycap.rr.com Thu, 08 Mar 2001 17:21:03 -0500
Date: Thu, 08 Mar 2001 17:21:03 -0500
From: Bill Lorensen wlorens1@nycap.rr.com
Subject: [Insight-developers] slice.h
I had already checked in that change. We had an update problem last evening on VC++.
Bill
At 01:30 PM 3/8/01 -0700, Joshua Cates wrote:
>Hi,
>
>Sounds like I should just #include then to conform to the
>standard. This has worked to define std::slice on VC++ in the past.
>
>Josh.
>______________________________
> Josh Cates
> School of Computer Science
> University of Utah
> Email: cates@cs.utah.edu
> Phone: (801) 587-7697
> URL: www.cs.utk.edu/~cates
>
>
>On Thu, 8 Mar 2001, Brad King wrote:
>
> > According to the C++ standard, the "slice" class is defined by
> > #include
> >
> > I don't know for sure if this works on all platforms, though.
> > -Brad
> >
> > At 11:48 AM 3/8/2001 -0700, Joshua Cates wrote:
> > >Hello,
> > >
> > >Where does slice.h reside on windows machines with VC++? I need a
> > >cross-platform way to #include "std/slice.h" (see todays dashboard under
> > >Win-NT-4.0-VC++)
> > >
> > >Thanks,
> > >
> > >Josh.
> >
> >
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
> >
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From lorensen@crd.ge.com Fri, 9 Mar 2001 09:00:08 -0500
Date: Fri, 9 Mar 2001 09:00:08 -0500
From: Lorensen, William E (CRD) lorensen@crd.ge.com
Subject: [Insight-developers] Registration Patterns
Folks,
Marc Levoy has published a paper that analyzes the ICP registration algorithm. He has borken it down
nicely and examined variations of each step in the process.
This may be of interest to us as to the approach he has taken. His analysis can map nicely into an OO
framework. Somwething similar may be appropriate for Insight's registration framework.
Bill
http://graphics.stanford.edu/papers/fasticp/
From ibanez@cs.unc.edu Fri, 09 Mar 2001 10:34:15 -0500
Date: Fri, 09 Mar 2001 10:34:15 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Registration Patterns
It is certainly a good breakdown of the ICP algorithm,
seems that we should be able implement it as a specialized
version of the registration framework in Insight.
Thanks,
Luis
"Lorensen, William E (CRD)" wrote:
> Folks,
>
> Marc Levoy has published a paper that analyzes the ICP registration algorithm. He has borken it down
> nicely and examined variations of each step in the process.
>
> This may be of interest to us as to the approach he has taken. His analysis can map nicely into an OO
> framework. Somwething similar may be appropriate for Insight's registration framework.
>
> Bill
>
> http://graphics.stanford.edu/papers/fasticp/
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
--
______________________________________________________________________
Luis Ibanez
Research Assistant Professor - Division of Neurosurgery
University of North Carolina at Chapel Hill
CB# 7060, Chapel Hill, NC 27599
email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
phone : (919)-843-9961 fax : (919)-966-6627
______________________________________________________________________
From gee@GRASP.cis.upenn.edu Fri, 9 Mar 2001 10:55:37 -0500
Date: Fri, 9 Mar 2001 10:55:37 -0500
From: James Gee gee@GRASP.cis.upenn.edu
Subject: [Insight-developers] Registration Patterns
i agree with luis. it's very nice but specialized enough that i don't see
it replacing the general framework that the group has already put together.
>
> It is certainly a good breakdown of the ICP algorithm,
> seems that we should be able implement it as a specialized
> version of the registration framework in Insight.
>
> Thanks,
>
> Luis
>
>
>
> "Lorensen, William E (CRD)" wrote:
>
> > Folks,
> >
> > Marc Levoy has published a paper that analyzes the ICP registration
algorithm. He has borken it down
> > nicely and examined variations of each step in the process.
> >
> > This may be of interest to us as to the approach he has taken. His
analysis can map nicely into an OO
> > framework. Somwething similar may be appropriate for Insight's
registration framework.
> >
> > Bill
> >
> > http://graphics.stanford.edu/papers/fasticp/
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
>
> --
> ______________________________________________________________________
>
> Luis Ibanez
> Research Assistant Professor - Division of Neurosurgery
> University of North Carolina at Chapel Hill
> CB# 7060, Chapel Hill, NC 27599
> email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
> phone : (919)-843-9961 fax : (919)-966-6627
> ______________________________________________________________________
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From vikram@insightful.com Fri, 9 Mar 2001 15:08:58 -0500
Date: Fri, 9 Mar 2001 15:08:58 -0500
From: vikram@insightful.com vikram@insightful.com
Subject: [Insight-developers] Re: Testing/2
Synopsis: Need a hook from the dashboard to the bug tracker
Responsible-Changed-From-To: bill->lorensen
Responsible-Changed-By: vikram
Responsible-Changed-When: Fri Mar 9 15:08:53 2001
Responsible-Changed-Why:
Bill is responsible for this now.
http://public.kitware.com/cgi-bin/gnatsweb.pl?cmd=view&pr=2&database=insight
From millerjv@crd.ge.com Fri, 9 Mar 2001 16:17:43 -0500
Date: Fri, 9 Mar 2001 16:17:43 -0500
From: millerjv@crd.ge.com millerjv@crd.ge.com
Subject: [Insight-developers] Re: Testing/2
Synopsis: Need a hook from the dashboard to the bug tracker
Responsible-Changed-From-To: lorensen->blezek
Responsible-Changed-By: millerjv
Responsible-Changed-When: Fri Mar 9 16:17:42 2001
Responsible-Changed-Why:
Dan writes all the dashboard XSLT.
http://public.kitware.com/cgi-bin/gnatsweb.pl?cmd=view&pr=2&database=insight
From vikram@insightful.com Fri, 9 Mar 2001 14:39:58 -0800 (PST)
Date: Fri, 9 Mar 2001 14:39:58 -0800 (PST)
From: Vikram Chalana x314 vikram@insightful.com
Subject: [Insight-developers] gnats ready to go
GNATS has been setup and is now ready to be used for filing ITK bugs.
All the people listed in UserList.txt have access to the gnats database
at:
http://public.kitware.com/cgi-bin/gnatsweb.pl
Use your email address (as listed in UserList.txt) as your password.
Four categories of bugs are currently defined (new categories can be
easily added as we move forward):
Build
Documentation
Source
Testing
Everyone on this email list is set to get a notification whenever a bug is
added or its status is changed. Again the notification list can be changed
as needed.
As the BUG-CZAR, I will be monitoring and cleaning the bug list
periodically.
Thanks,
Vikram
From ibanez@cs.unc.edu Sat, 10 Mar 2001 11:22:27 -0500
Date: Sat, 10 Mar 2001 11:22:27 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Problems with CMake
Hi,
I just updated from the repository
and when I run CMake it is reporting errors.
a small window pops up with the message
CMake
error:
and the OK button.
It appears several times.
Any ideas ?
Thanks
Luis
From wlorens1@nycap.rr.com Sat, 10 Mar 2001 11:27:37 -0500
Date: Sat, 10 Mar 2001 11:27:37 -0500
From: Bill Lorensen wlorens1@nycap.rr.com
Subject: [Insight-developers] Problems with CMake
Luis,
You may need to run CMakeSetup
At 11:22 AM 3/10/01 -0500, Luis Ibanez wrote:
>Hi,
>
>I just updated from the repository
>and when I run CMake it is reporting errors.
>a small window pops up with the message
>
>CMake
>error:
>
>and the OK button.
>
>It appears several times.
>
>Any ideas ?
>
>Thanks
>
>Luis
>
>
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From ibanez@cs.unc.edu Sat, 10 Mar 2001 11:52:36 -0500
Date: Sat, 10 Mar 2001 11:52:36 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Problems with CMake
Sorry, I forgot to tell that this is happening on WinNT,
I run the executable in the CMakeSetup project, but it
still doing the same.
It is unfortunate that the message is not very informative...
Parsing the repository I saw that CMakeList.txt files
should have an ending new line, could that be related ?
Should I make a clean checkout ?
Thanks
Luis
-------
Bill Lorensen wrote:
> Luis,
> You may need to run CMakeSetup
>
> At 11:22 AM 3/10/01 -0500, Luis Ibanez wrote:
> >Hi,
> >
> >I just updated from the repository
> >and when I run CMake it is reporting errors.
> >a small window pops up with the message
> >
> >CMake
> >error:
> >
> >and the OK button.
> >
> >It appears several times.
> >
> >Any ideas ?
> >
> >Thanks
> >
> >Luis
> >
> >
> >
> >
> >
> >_______________________________________________
> >Insight-developers mailing list
> >Insight-developers@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/insight-developers
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
--
______________________________________________________________________
Luis Ibanez
Research Assistant Professor - Division of Neurosurgery
University of North Carolina at Chapel Hill
CB# 7060, Chapel Hill, NC 27599
email : ibanez@cs.unc.edu home : http://www.cs.unc.edu/~ibanez
phone : (919)-843-9961 fax : (919)-966-6627
______________________________________________________________________
From ibanez@cs.unc.edu Sat, 10 Mar 2001 11:58:38 -0500
Date: Sat, 10 Mar 2001 11:58:38 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkIndex initialization
Hi,
I just noticed that itkIndex has a static variable
initialized to zero.
So, Instead of writting
itk::Index<3> start;
start[0] = 0;
start[1] = 0;
start[2] = 0;
itk::Region<3> region;
region.SetIndex( start );
we can write:
itk::Region<3> region;
region.SetIndex( itk::Index<3>::ZeroIndex );
That reduces a little the protocol for initializing
images.
Luis
From bill.hoffman@kitware.com Sat, 10 Mar 2001 12:20:10 -0500
Date: Sat, 10 Mar 2001 12:20:10 -0500
From: William A. Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] Problems with CMake
That is a useful error isn't it? The problem is caused because you need
to do a cvs update -d, Brad has added several new directories, and for some
reason the new CMake commands write out some empty error messages.
-Bill
At 11:22 AM 3/10/01 -0500, Luis Ibanez wrote:
>Hi,
>
>I just updated from the repository
>and when I run CMake it is reporting errors.
>a small window pops up with the message
>
>CMake
>error:
>
>and the OK button.
>
>It appears several times.
>
>Any ideas ?
>
>Thanks
>
>Luis
>
>
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From chenting@graphics.cis.upenn.edu Sat, 10 Mar 2001 17:00:43 -0600
Date: Sat, 10 Mar 2001 17:00:43 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] about testing
Hi! I checked in 2 test program, itkDeformableTest and itkGibbsTest. Both of
them are running ok on my system (NT4.0 VC++6.0), my pal at Columbia also
checked out from the CVS server and the program runs well there, too.
however, on the dashboard, the status of my program is still not run, can
guys from kitware or GE check out what is run there for me? ( for
itkGibbsTest, the problem is that I used a local file, but there is no
reason for itkDeformableTest )
Thanks a lot!
ting
From bill.hoffman@kitware.com Sat, 10 Mar 2001 17:56:50 -0500
Date: Sat, 10 Mar 2001 17:56:50 -0500
From: William A. Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] about testing
Your programs have a getchar() call in them. While this is fine
for interactive testing, there is no there to press a key at night!
At 05:00 PM 3/10/01 -0600, Ting Chen wrote:
>Hi! I checked in 2 test program, itkDeformableTest and itkGibbsTest. Both of
>them are running ok on my system (NT4.0 VC++6.0), my pal at Columbia also
>checked out from the CVS server and the program runs well there, too.
>however, on the dashboard, the status of my program is still not run, can
>guys from kitware or GE check out what is run there for me? ( for
>itkGibbsTest, the problem is that I used a local file, but there is no
>reason for itkDeformableTest )
>Thanks a lot!
>ting
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From brad.king@kitware.com Mon, 12 Mar 2001 09:41:55 -0500 (EST)
Date: Mon, 12 Mar 2001 09:41:55 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Problems with CMake
Luis,
The problem is not immediately obvious from my new commands. I have added
code to cmCommand that should produce a more descriptive error message in
the case of an empty message from a GetError() call. If you can still
duplicate the problem, do a cvs update and try it again.
Thanks,
-Brad
On Sat, 10 Mar 2001, William A. Hoffman wrote:
> That is a useful error isn't it? The problem is caused because you need
> to do a cvs update -d, Brad has added several new directories, and for some
> reason the new CMake commands write out some empty error messages.
>
> -Bill
>
>
> At 11:22 AM 3/10/01 -0500, Luis Ibanez wrote:
> >Hi,
> >
> >I just updated from the repository
> >and when I run CMake it is reporting errors.
> >a small window pops up with the message
> >
> >CMake
> >error:
> >
> >and the OK button.
> >
> >It appears several times.
> >
> >Any ideas ?
> >
> >Thanks
> >
> >Luis
From ibanez@cs.unc.edu Mon, 12 Mar 2001 12:08:05 -0500 (EST)
Date: Mon, 12 Mar 2001 12:08:05 -0500 (EST)
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Problems with CMake
Hi,
I did an update with -d option
as Bill suggested, and it is
working fine right now.
Now, I put that on the .cvsrc file,
as default options: "update -d -P"
Thanks,
Luis
---------------------
On Mon, 12 Mar 2001, Brad King wrote:
> Luis,
>
> The problem is not immediately obvious from my new commands. I have added
> code to cmCommand that should produce a more descriptive error message in
> the case of an empty message from a GetError() call. If you can still
> duplicate the problem, do a cvs update and try it again.
>
> Thanks,
> -Brad
>
> On Sat, 10 Mar 2001, William A. Hoffman wrote:
>
> > That is a useful error isn't it? The problem is caused because you need
> > to do a cvs update -d, Brad has added several new directories, and for some
> > reason the new CMake commands write out some empty error messages.
> >
> > -Bill
> >
> >
> > At 11:22 AM 3/10/01 -0500, Luis Ibanez wrote:
> > >Hi,
> > >
> > >I just updated from the repository
> > >and when I run CMake it is reporting errors.
> > >a small window pops up with the message
> > >
> > >CMake
> > >error:
> > >
> > >and the OK button.
> > >
> > >It appears several times.
> > >
> > >Any ideas ?
> > >
> > >Thanks
> > >
> > >Luis
>
>
From vikram@insightful.com Mon, 12 Mar 2001 19:04:34 -0500
Date: Mon, 12 Mar 2001 19:04:34 -0500
From: vikram@insightful.com vikram@insightful.com
Subject: [Insight-developers] Re: Documentation/8
Synopsis: style guide needs to become a real document
State-Changed-From-To: open->closed
State-Changed-By: vikram
State-Changed-When: Mon Mar 12 19:04:34 2001
State-Changed-Why:
duplicated.
http://public.kitware.com/cgi-bin/gnatsweb.pl?cmd=view&pr=8&database=insight
From millerjv@crd.ge.com Tue, 13 Mar 2001 09:16:29 -0500
Date: Tue, 13 Mar 2001 09:16:29 -0500
From: Miller, James V (CRD) millerjv@crd.ge.com
Subject: [Insight-developers] Preparation for internal alpha
In preparation for the internal alpha of insight, I propose that we compose a list of capabilities
that will be ready for inclusion in the toolkit by March 31, 2001. There is a lot of collaboration
work between various contractors, but some people are waiting for some of the higher level frameworks
(registration, level sets, region growing) to be completed prior to converting some of their existing
(insight) implementations to the new frameworks. What I'd like to see is a list of dates for when
people think these various frameworks may be available. Since this first alpha is an "internal"
alpha, it is really meant to timestamp the repository and bring independent development paths back
together. I think it is reasonable to strive to have the APIs for these frameworks in place for the
internal alpha but not have converted everything over to the new API. Development will continue
furiously through the June public alpha (at which time most of the existing techniques should be
converted to the new frameworks).
I have added an html file to the cvs repository to collect these schedules
(HTMLPrivate/InternalAlphaSchedule.html), please edit it and place in your timelines. You can add a
row to the table for each capability/algorithm you'll be providing. (I just put in one item to start
with). There is a link from the main Insight page to the schedule at
http://public.kitware.com/Insight/HTMLPrivate/InternalAlphaSchedule.html
I believe that we decided in Utah not to use GNATS for project management, but rather limit GNATS use
to actual bugs. However, maintaining this HTML file for schedules and todo items is going to get
rather cumbersome. Does anyone know of any "free" web based todo list managers? SourceForge.net
would be nice but I don't think we are ready to publicize our project yet.
Of course, we also need to drive down all the compiler errors and warnings prior to the internal
alpha release!
Jim Miller
_____________________________________
Computer Graphics & Systems Program
GE Corporate Research & Development
Bldg. KW, Room C218B
P.O. Box 8, Schenectady NY 12301
millerjv@crd.ge.com
(518) 387-4005, Dial Comm: 8*833-4005,
Cell: (518) 505-7065, Fax: (518) 387-6981
<>
begin 600 James Miller (E-mail).vcf
M0D5'24XZ5D-!4D0-"E9%4E-)3TXZ,BXQ#0I..DUI;&QE7-T96US(%!R;V=R86T[0DQ$1R!+5RP@4DT@0S(Q.$(]
M,$0],$%03R!";W@@.#M38VAE;F5C/0T*=&%D>3M.63LQ,C,P,3M5;FET960@
M4W1A=&5S(&]F($%M97)I8V$-"DQ!0D5,.U=/4DL[14Y#3T1)3D<]455/5$5$
M+5!224Y404),13I#;VUP=71E2P@3ED@,3(S,#$],$0],$%5;FET960@4W1A=&5S(&]F
M($%M97)I8V$-"D5-04E,.U!2148[24Y415).150Z;6EL;&5R:G9`8W)D+F=E
G+F-O;0T*4D56.C(P,#`P,3(X5#$V,C8U-EH-"D5.1#I60T%21`T*
`
end
From hughett@mercur.uphs.upenn.edu Tue, 13 Mar 2001 10:48:11 -0500
Date: Tue, 13 Mar 2001 10:48:11 -0500
From: Paul Hughett hughett@mercur.uphs.upenn.edu
Subject: [Insight-developers] Build broken on Linux 6.2
I just updated to the latest ITK version from the repository, and did
"configure", "make clean", and "make". I got the following fatal error
message almost immediately:
make[3]: Leaving directory `/home/hughett/work/Insight/Utilities/Cable'
cd Cable; make -w all
make[3]: Entering directory `/home/hughett/work/Insight/Utilities/Cable'
make[3]: *** No rule to make target `sourceParser.o', needed by `generateWrappers'. Stop.
Paul Hughett
From brad.king@kitware.com Tue, 13 Mar 2001 13:48:10 -0500 (EST)
Date: Tue, 13 Mar 2001 13:48:10 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Build broken on Linux 6.2
> I just updated to the latest ITK version from the repository, and did
> "configure", "make clean", and "make". I got the following fatal error
> message almost immediately:
>
> make[3]: Leaving directory `/home/hughett/work/Insight/Utilities/Cable'
> cd Cable; make -w all
> make[3]: Entering directory `/home/hughett/work/Insight/Utilities/Cable'
> make[3]: *** No rule to make target `sourceParser.o', needed by `generateWrappers'. Stop.
Did you do a "cvs update -d"?
-Brad
From hughett@mercur.uphs.upenn.edu Tue, 13 Mar 2001 13:50:26 -0500
Date: Tue, 13 Mar 2001 13:50:26 -0500
From: Paul Hughett hughett@mercur.uphs.upenn.edu
Subject: [Insight-developers] Build broken on Linux 6.2
> Did you do a "cvs update -d"?
Yes. And I just did it once again just to make sure that I wan't
misremembering it. Same problem.
Paul Hugehtt
From brad.king@kitware.com Tue, 13 Mar 2001 14:14:51 -0500 (EST)
Date: Tue, 13 Mar 2001 14:14:51 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Build broken on Linux 6.2
> I just updated to the latest ITK version from the repository, and did
> "configure", "make clean", and "make". I got the following fatal error
> message almost immediately:
>
> make[3]: Leaving directory `/home/hughett/work/Insight/Utilities/Cable'
> cd Cable; make -w all
> make[3]: Entering directory `/home/hughett/work/Insight/Utilities/Cable'
> make[3]: *** No rule to make target `sourceParser.o', needed by `generateWrappers'. Stop.
The problem should be fixed now. There was a stray makefile left in the
repository from my original development version of Cable before it was
integrated in to the Insight/Utilities directory and built with CMake.
It was confusing the in-place build process.
-Brad
From hughett@mercur.uphs.upenn.edu Tue, 13 Mar 2001 14:16:32 -0500
Date: Tue, 13 Mar 2001 14:16:32 -0500
From: Paul Hughett hughett@mercur.uphs.upenn.edu
Subject: [Insight-developers] Build broken on Linux 6.2
> The problem should be fixed now. There was a stray makefile left in the
> repository from my original development version of Cable before it was
> integrated in to the Insight/Utilities directory and built with CMake.
> It was confusing the in-place build process.
Yup, that fixed it. Thanks.
Paul Hughett
From brad.king@kitware.com Tue, 13 Mar 2001 14:25:57 -0500 (EST)
Date: Tue, 13 Mar 2001 14:25:57 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] itk::Mesh and Cell template parameters
Hello, all:
Those of you working on development related to the Mesh class and its
Cells should note that I have renamed some template parameters. The
second template parameter of the Mesh and Cell classes were called
"MeshType" and "CellType", but the classes that are passed to them are
simply full of typedefs and enums. They play the role of "trait" classes,
so I have renamed the template parameters to "MeshTraits" and
"CellTraits". The default trait classes "itk::DefaultStaticMeshType" and
"itk::DefaultDynamicMeshType" have been renamed accordingly.
If anyone has any code that used the old names (that isn't checked in),
you will have to update it to work with the new names. Also, anyone who
has created additional trait classes should rename them accordingly as
well (MyMeshType -> MyMeshTraits).
-Brad
From ibanez@cs.unc.edu Tue, 13 Mar 2001 14:52:28 -0500
Date: Tue, 13 Mar 2001 14:52:28 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] gdb with Cygwin
Hi,
I recently downloaded Cygwin, and there are
a lot of improvements in the current version.
The set of tools is now extended at the point
that you really feel like in Linux.
One of the nicest things, is the version of gdb,
it has a graphic interface and a very intuitive
access to breakpoints, stack, registers, mix
of source and assembler code, and view of data.
Luis
From brad.king@kitware.com Tue, 13 Mar 2001 15:30:10 -0500 (EST)
Date: Tue, 13 Mar 2001 15:30:10 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Coding style, MSVC INTERNAL COMPILER ERRORs, and enums
Hello, all:
I just tracked down another ICE by MSVC related to the bug in it that I
previously reported to the list. The fix has been checked in, but we can
avoid the issue altogether with a little care in the style of our code.
Here is a typical example that causes the problem:
template
class MyClass
{
void MyMethod(typename TInputImage::PixelType pixel);
};
//...and through some chain of includes when MyClass is used...
template <
unsigned int VDimension
> // MSVC fails here with an ICE
class SomeOtherClass { /*...*/ };
The only explanation for this problem is a bug in MSVC's compiler. The
solution is simple, and should be used anyway just for readability. Any
type that is qualified inside another type, namespace, or is a typename
type should not be directly used as a method argument. Instead, typedef
the type and use the new name as the argument. Here is the example above
with this change:
template
class MyClass
{
typedef typename TInputImage::PixelType InputPixelType;
void MyMethod(InputPixelType pixel);
};
// MSVC will no longer ICE on a "template " block.
It would be good if everyone could use this style, both for the ICE
problem, and for readability. The same thing should be done with enum
values as well (like ImageDimension). In fact, it is necessary to do this
with enums since the compiler does not get the "typename" hint for an
enum value.
Earlier today I spent an hour tracking down such an enum problem in
itk::DanielssonDistanceMapImageFilter. Here is a simplified example of
the problem:
template struct X {};
template struct A
{
void Method(X);
};
template
void A::Method(X) {}
GCC gives the following errors while parsing this code:
prototype for `void A::Method(X)' does not match any in
class `A'
candidate is: void A::Method(X)
In method `void A::Method(X)':
template definition of non-template `void A::Method(X)'
Note that the candidate it gives consists of the EXACT SAME text as the
prototype that the candidate does not match! The solution to this problem
is to make sure the compiler knows that the value is an enum.
template struct X {};
template struct A
{
enum { EnumValue = T::EnumValue };
void Method(X);
};
template
void A::Method(X) {}
This is actually a C++ issue not related to GCC or any other compiler (I
don't have a quote from the C++ standard this time, though). Again, it
can be fixed with good style in our code.
-Brad
From brad.king@kitware.com Tue, 13 Mar 2001 16:58:44 -0500 (EST)
Date: Tue, 13 Mar 2001 16:58:44 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] typename platform conflict
Hello, all:
Well, at long last we have encountered a situation in which the
intersection of the capabilities of all our compilers is the empty
set. We will have to use a macro to resolve this one.
The problem occurs when a typename type is used in a default template
argument:
typename
class X {};
SGI's MIPSpro will not accept this code without the typename keyword.
MSVC will not accept this code unless the typename keyword is missing.
I have added a macro definition to itkWin32Header.h:
#define TYPENAME typename
in the case of MSVC, the macro will be defined as
#define TYPENAME
The new code for the above example is
typename
class X {};
This macro should ONLY be used for default template arguments. It should
NEVER be used in place of plain old "typename" in other situations. If
anyone has another problem for which you think you need the TYPENAME
macro, post it to this list first for discussion.
Also, now would be a good time to discuss the name of this
macro. TYPENAME is minimally disruptive to reading the code, but may
mislead people into using it when it should not be used (not to mention
name conflicts with other libraries). We might want to consider
ITK_DEFAULT_ARGUMENT_TYPENAME instead. Anyone have thoughts on this?
-Brad
From millerjv@crd.ge.com Tue, 13 Mar 2001 17:53:24 -0500
Date: Tue, 13 Mar 2001 17:53:24 -0500
From: Miller, James V (CRD) millerjv@crd.ge.com
Subject: [Insight-developers] typename platform conflict
Can we solve this by inserting a dummy class, so the code would look like
template
class DummyTrait
{
typedef T::SomeType SomeType;
}
typename::SomeType>
class X{}
instead of using a macro. (Can you tell I hate macros....)
-----Original Message-----
From: Brad King [mailto:brad.king@kitware.com]
Sent: Tuesday, March 13, 2001 4:59 PM
To: Insight Developers
Subject: [Insight-developers] typename platform conflict
Hello, all:
Well, at long last we have encountered a situation in which the
intersection of the capabilities of all our compilers is the empty
set. We will have to use a macro to resolve this one.
The problem occurs when a typename type is used in a default template
argument:
typename
class X {};
SGI's MIPSpro will not accept this code without the typename keyword.
MSVC will not accept this code unless the typename keyword is missing.
I have added a macro definition to itkWin32Header.h:
#define TYPENAME typename
in the case of MSVC, the macro will be defined as
#define TYPENAME
The new code for the above example is
typename
class X {};
This macro should ONLY be used for default template arguments. It should
NEVER be used in place of plain old "typename" in other situations. If
anyone has another problem for which you think you need the TYPENAME
macro, post it to this list first for discussion.
Also, now would be a good time to discuss the name of this
macro. TYPENAME is minimally disruptive to reading the code, but may
mislead people into using it when it should not be used (not to mention
name conflicts with other libraries). We might want to consider
ITK_DEFAULT_ARGUMENT_TYPENAME instead. Anyone have thoughts on this?
-Brad
_______________________________________________
Insight-developers mailing list
Insight-developers@public.kitware.com
http://public.kitware.com/mailman/listinfo/insight-developers
From brad.king@kitware.com Tue, 13 Mar 2001 18:10:26 -0500 (EST)
Date: Tue, 13 Mar 2001 18:10:26 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] typename platform conflict
Jim,
> Can we solve this by inserting a dummy class, so the code would look like
>
> template
> class DummyTrait
> {
> typedef T::SomeType SomeType;
> }
>
> typename::SomeType>
> class X{}
>
> instead of using a macro. (Can you tell I hate macros....)
Sorry, but no. This goes back to the reason the typename keyword exists
in the first place:
template
struct DummyTrait
{
typedef T::SomeType SomeType;
};
template ::SomeType>
class X {};
// .....
template <>
struct DummyTrait
{
enum { SomeType = 1 };
};
// .....
template class X; // uh-oh!
Obviously we wouldn't ever write the specialization of DummyTrait, but the
compiler doesn't know if we might do this or not when it parses
DummyTrait::SomeType. The typename keyword exists to enforce that
SomeType will be a type.
I'm quite confident that the macro is the only solution, but at least it
is a somewhat clean macro. If you want to see nasty macro usage, look at
BOOST classes. They have code that looks something like:
template
class X {};
so that they can remove the use of default arguments on compilers that
don't support them as well as adjust the use of the typename keyword.
-Brad
From chenting@graphics.cis.upenn.edu Tue, 13 Mar 2001 18:15:02 -0600
Date: Tue, 13 Mar 2001 18:15:02 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] itk::Mesh and Cell template parameters
Hi! Brad, although you must have plenty of reasons to make this kind of
modification, I still doubt is it ok to change the structure name this
often, especially when the modifications make not too much sense.
ting
-----Original Message-----
From: Brad King
To: Insight Developers
Date: Tuesday, March 13, 2001 1:23 PM
Subject: [Insight-developers] itk::Mesh and Cell template parameters
>Hello, all:
>
>Those of you working on development related to the Mesh class and its
>Cells should note that I have renamed some template parameters. The
>second template parameter of the Mesh and Cell classes were called
>"MeshType" and "CellType", but the classes that are passed to them are
>simply full of typedefs and enums. They play the role of "trait" classes,
>so I have renamed the template parameters to "MeshTraits" and
>"CellTraits". The default trait classes "itk::DefaultStaticMeshType" and
>"itk::DefaultDynamicMeshType" have been renamed accordingly.
>
>If anyone has any code that used the old names (that isn't checked in),
>you will have to update it to work with the new names. Also, anyone who
>has created additional trait classes should rename them accordingly as
>well (MyMeshType -> MyMeshTraits).
>
>-Brad
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From brad.king@kitware.com Wed, 14 Mar 2001 10:27:53 -0500 (EST)
Date: Wed, 14 Mar 2001 10:27:53 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] itk::Mesh and Cell template parameters
Ting,
> Hi! Brad, although you must have plenty of reasons to make this kind
> of modification, I still doubt is it ok to change the structure name
> this often, especially when the modifications make not too much sense.
I probably should have proposed the change to the list before making
it. In the early development last summer, this wouldn't have mattered,
but such changes affect many more people now. Sorry about that.
Anyway, what doesn't make sense to you about the new name? The old one
was a misnomer from the beginning. A template parameter whose only
purpose is to encapsulate typedefs and other things that could be template
parameters themselves is generally considered a set of "traits".
-Brad
From yj76@columbia.edu Wed, 14 Mar 2001 14:40:52 -0500
Date: Wed, 14 Mar 2001 14:40:52 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] Depth of nested templates
As Brad told me, I have this too deep nested template problem in my code:
typedef itk::Point EdgeInfo;
std::vector< std::deque > rawEdges;
I don't know how far I can go. the first line should be fine since it is
just an instantiation of itkPoint.
can we go std::vector or std::deque?
any suggestion on this?
the problem is I have a list of polygon (size of the list is unkown), for
each polygon there will be a list of EdgeInfo (again the size of those lists
are unknow).
what if I know the size of the polygon list as an nonconstant varible.
are we allow to do this:
typedef std::deque EdgeInfoQ
EdgeInfoQ *rawEdges = new EdgeInfoQ[knowSize];
I remember somebody addressed this standard C++ "new" operation issue
before, forgot the details.
Yinpeng
From brad.king@kitware.com Wed, 14 Mar 2001 15:23:55 -0500 (EST)
Date: Wed, 14 Mar 2001 15:23:55 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Depth of nested templates
> As Brad told me, I have this too deep nested template problem in my code:
> [.....]
Just so that everyone is clear on what I mean by nested template depth,
here is an example:
// Recursive type definition.
template
struct MyStruct
{
typedef typename MyStruct::Type Type;
};
// End the recursion at zero.
template <>
struct MyStruct<0>
{
typedef int Type;
};
Now, to determine the type of "MyStruct<10>::Type", the compiler has to
instantiate "MyStruct<10>", "MyStruct<9>", ..., "MyStruct<1>" in order to
get to "MyStruct<0>::Type". This is a depth of 10 template
instantiations. Compilers like GCC have a limit to the depth they will
allow. By default it is 17 for GCC. The reason a limit is necessary is
so that the compiler doesn't get into an infinity loop of instantiations
by, say, trying to find the type "MyStruct<-1>::Type".
On GCC, the limit can be increased with the "-f-template-depth-N" flag,
where N is the depth limit. MSVC doesn't seem to have a limit, and just
crashes. If you want, cut and paste the code at the bottom of this
message into a new file in Visual Studio's editor, and watch what happens.
Make sure everything is saved first :)
What Yinpeng has encountered is code that naturally reaches the 17
instantiation limit. It gives an error on GCC unless you specify
something like -f-template-depth-50 (I didn't actually find the exact
number needed). We could consider requiring this flag to build Insight,
but if there are other, just as clean implementations, then they should be
used instead. Deep template instantiation depths are slow to build
anyway. Thoughts, anyone?
-Brad
--------------------------
template
struct MyStruct
{
typedef typename MyStruct::Type Type;
};
MyStruct<10>::Type x;
From will.schroeder@kitware.com Wed, 14 Mar 2001 15:42:18 -0500
Date: Wed, 14 Mar 2001 15:42:18 -0500
From: Will Schroeder will.schroeder@kitware.com
Subject: [Insight-developers] Depth of nested templates
Hi Folks-
Looking at the Voronoi class, I can't help but wonder why this is not a filter? It's a subclass of Mesh?? If it is to exist in the data processing pipeline, the algorithm should be separated from its input and output...algorithm should be an eventual subclass of ProcessObject and the input and output should be subclasses of DataObject. (I'm not picking on Yinpeng, there are other classes that could be filters that are not.)
I think that we are not thinking in pipeline fashion. I think that it is essential that the use and interface to classes is as uniform as possible.
Will
At 02:40 PM 3/14/2001 -0500, Yinpeng Jin wrote:
>As Brad told me, I have this too deep nested template problem in my code:
>
>typedef itk::Point EdgeInfo;
>std::vector< std::deque > rawEdges;
>
>I don't know how far I can go. the first line should be fine since it is
>just an instantiation of itkPoint.
>can we go std::vector or std::deque?
>any suggestion on this?
>the problem is I have a list of polygon (size of the list is unkown), for
>each polygon there will be a list of EdgeInfo (again the size of those lists
>are unknow).
>
>what if I know the size of the polygon list as an nonconstant varible.
>are we allow to do this:
>
>typedef std::deque EdgeInfoQ
>EdgeInfoQ *rawEdges = new EdgeInfoQ[knowSize];
>
>I remember somebody addressed this standard C++ "new" operation issue
>before, forgot the details.
>
>Yinpeng
>
>
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From yj76@columbia.edu Wed, 14 Mar 2001 15:57:40 -0500
Date: Wed, 14 Mar 2001 15:57:40 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] Depth of nested templates
yes, I agree, it should be a MeshToMeshFilter, which take input as mesh of
vertexCell and output as mesh of polygon.
but this is just a part of my segmentation class, i just want to keep it as
simple as possible. There is no point for me (for this particular algorithm)
to save the vertex into a mesh because I will be using them intensively, and
I feel it is more straighforward to just save the vertices into a
vector than into a mesh of vertexCell. otherwise, in order to
generate the Voronoi Diagram, I need to first generate a Mesh of vertex.
Sorry that I didn't take too much into accout of the whole architecture
orgnization, and quite frankly, I think our whole segmenation groups cannot
take care of this issue too much, since we didn't take part into the
development of the basic architecture and we are not really familar with
this issue.
So maybe in our future development, at least for our segmentation group, we
need to write down a brief proposal about what we are going to do, and the
architecture people can guide us with proper framework.
thanks,
Yinpeng
----- Original Message -----
From: "Will Schroeder"
To: "Yinpeng Jin" ; "Insight Developers"
Sent: Wednesday, March 14, 2001 3:42 PM
Subject: Re: [Insight-developers] Depth of nested templates
> Hi Folks-
>
> Looking at the Voronoi class, I can't help but wonder why this is not a
filter? It's a subclass of Mesh?? If it is to exist in the data processing
pipeline, the algorithm should be separated from its input and
output...algorithm should be an eventual subclass of ProcessObject and the
input and output should be subclasses of DataObject. (I'm not picking on
Yinpeng, there are other classes that could be filters that are not.)
>
> I think that we are not thinking in pipeline fashion. I think that it is
essential that the use and interface to classes is as uniform as possible.
>
> Will
>
>
>
> At 02:40 PM 3/14/2001 -0500, Yinpeng Jin wrote:
> >As Brad told me, I have this too deep nested template problem in my code:
> >
> >typedef itk::Point EdgeInfo;
> >std::vector< std::deque > rawEdges;
> >
> >I don't know how far I can go. the first line should be fine since it is
> >just an instantiation of itkPoint.
> >can we go std::vector or std::deque?
> >any suggestion on this?
> >the problem is I have a list of polygon (size of the list is unkown), for
> >each polygon there will be a list of EdgeInfo (again the size of those
lists
> >are unknow).
> >
> >what if I know the size of the polygon list as an nonconstant varible.
> >are we allow to do this:
> >
> >typedef std::deque EdgeInfoQ
> >EdgeInfoQ *rawEdges = new EdgeInfoQ[knowSize];
> >
> >I remember somebody addressed this standard C++ "new" operation issue
> >before, forgot the details.
> >
> >Yinpeng
> >
> >
> >
> >
> >
> >_______________________________________________
> >Insight-developers mailing list
> >Insight-developers@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/insight-developers
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
From will.schroeder@kitware.com Thu, 15 Mar 2001 11:36:04 -0500
Date: Thu, 15 Mar 2001 11:36:04 -0500
From: Will Schroeder will.schroeder@kitware.com
Subject: [Insight-developers] (no subject)
Hi Folks-
In my wanderings through the dashboard and into the source code it appears that DeformableMesh needs some serious work. It was checked in about two weeks ago and the errors it is generating have yet to be addressed. The problems include:
* Style
- not using m_ as in m_Resolution
- indentation is whacked out
* Documentation
- completely wrong, copied from FilterMeshToMesh (an old name anyway)
* Template parameters are bad...it's only using one of the to defined in itkMesh, which of course limits the classes flexibility and usefulness
* The use of pow() is causing compile errors because the types are ambiguous (i.e., an argument to the pow() may be integer...it needs to be cast to either float or double depending on the type of the first argument.
Please clean it up. Thanks.
Will
From ibanez@cs.unc.edu Thu, 15 Mar 2001 12:37:28 -0500
Date: Thu, 15 Mar 2001 12:37:28 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Image/ImageAdaptor/DataAccessor modification
Hi,
Relations between ImageAdaptor, Image and DataAccessor
have slightly changed.
Before a DataAccessor was only a type with static methods
defined in Image and ImageAdaptor. Iterators used it to
convert types during pixel access.
Now a member variable of DataAccessor type is declared
in Image and ImageAdaptor. In this way, the Accessor can have
some state information.
This is used to make possible a N-th element adaptor.
itkNthElementImageAdaptor
itkNthElementDataAccessor
Given and image whose pixels are of type container:
itkVector, itkPoint, std::vector... and in general, any type
that provides a operator[](unsigned int) method.
The image can be adapted to make it look like a scalar
image whose pixels are the n-th elements of the adapted
image.
The selection of the element number can be done at run time
using the SelectNthElement() method in NthElementImageAdaptor.
+--------------+
| | Here it looks
+----+ N-th | like a scalar
+---------------------------+ | element | image. Each pixel
| Image, 3> |->| Image | is the n-th element
+---------------------------+ | Adaptor | of the corresponding
+----+ | pixel in the adapted
| m_ElementNum |
+--------------+
---
Pixel access is supposed to be inlined.
That seems to be the case in VC++ and gcc,
...but if you notice any big change in performance
on pixel access, I'll double check the iterator's
Get/Set method.
Thanks,
Luis
From yj76@columbia.edu Thu, 15 Mar 2001 13:32:34 -0500
Date: Thu, 15 Mar 2001 13:32:34 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] TCON on friday
what should I do if I want to attend the tcon? just dial in that number and
code? should I sign-up for something in advance?
BTW, please confirm that there is one tomorrow, since some of my group may
sign in and discuss our current project with the Architecture persons.
thanks,
yinpeng
From cates@cayenne.cs.utah.edu Thu, 15 Mar 2001 16:18:15 -0700 (MST)
Date: Thu, 15 Mar 2001 16:18:15 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] Image/ImageAdaptor/DataAccessor modification
Hi all, Luis,
I have run into a problem this week that we have been unable to resolve in
supporting the image adaptors for non-scalar data types. The problem is a
seemingly unavoidable penalty in working with non-scalar data. The penalty
results because manipulation of non-scalar data cannot be optimized down
to access by reference in many cases.
In trivial cases of scalar Get and Set, the compiler should optimize Get()
and Set() back down to an operator* call on the actual data pointer,
resulting in data access by reference and no loss in performance. But in
order to modify fields within more complex data types, you have to work
with the copy of the data that Get() returns and then copy back into the
memory location with Set(). (see the examples below)
I wrote a simple test program (shown below) which gave the following
typical results:
(1) gcc version 2.95.1 -O2 (Linux)
Speed of adaptor supporting interator (for reference) 80000
Speed of interator that does not support adaptors (for reference) 30000
a. Modifying scalar image using adaptor api... 110000
compensated = 30000
b. Modifying scalar image bypassing adaptor api... 50000
compensated = 20000
c. Modifying vector image using adaptor api... 1350000
compensated = 1270000
d. Modifying vector image bypassing adaptor api... 130000
compensated = 100000
(2) MIPSpro Compilers: Version 7.3.1.1m -O2 (IRIX)
Speed of adaptor supporting interator (for reference) 70000
Speed of interator that does not support adaptors (for reference) 20000
a. Modifying scalar image using adaptor api... 110000
compensated = 40000
b. Modifying scalar image bypassing adaptor api... 60000
compensated = 40000
c. Modifying vector image using adaptor api... 2090000
compensated = 2020000
d.Modifying vector image bypassing adaptor api... 170000
compensated = 150000
I have attempted to compensate for the difference in speed of the
different iterators. The correlation between iterator speed and adaptor
support is accidental and does not imply causality.
The compensated values for a and b show no significant difference as we
would expect. But compensated differences between c and d are significant
in both case 1 and 2.
The problem I've described imposes a serious penalty for certain cases and
merits more testing. Is there a solution within the adaptor framework that
I have not though of? or do we need an API to bypass adaptors in cases
where the trade-off between adaptor benefits and penalties is not
acceptable?
Let's spend some time discussing adaptor issues in the TCON tomorrow.
Particularly I'd like to hear other developers' experiences with
supporting data accessors in their algorithms--other issues they have
encountered or resolved--and hear how folks are using image adaptors.
Josh.
-------------------- itkAdaptorComparisonTest.cxx
#include
#include
#include "itkImage.h"
#include "itkSimpleImageRegionIterator.h"
#include "itkImageRegionIterator.h"
#include "itkVector.h"
void AdaptorSupportedIteratorSpeed(itk::Image *img)
{
itk::SimpleImageRegionIterator >
it (img, img->GetRequestedRegion());
it.Begin();
while( ! it.IsAtEnd() )
{
++it;
}
}
void NoAdaptorSupportIteratorSpeed(itk::Image *img)
{
itk::ImageRegionIterator
it (img, img->GetRequestedRegion());
it.Begin();
while( ! it.IsAtEnd() )
{
++it;
}
}
void AdaptorSupportedModifyScalars(itk::Image *img)
{
itk::SimpleImageRegionIterator >
it (img, img->GetRequestedRegion());
it.Begin();
while( ! it.IsAtEnd() )
{
// *it += 3.435f;
it.Set(it.Get() + 3.435f);
++it;
}
}
void NoAdaptorSupportModifyScalars(itk::Image *img)
{
itk::ImageRegionIterator
it (img, img->GetRequestedRegion());
it.Begin();
while( ! it.IsAtEnd() )
{
*it += 3.435f;
++it;
}
}
void AdaptorSupportedModifyVectors(itk::Image, 3>
*img)
{
typedef itk::Vector VectorType;
const unsigned int N = 3;
unsigned int i;
VectorType temp_vector;
itk::SimpleImageRegionIterator >
it (img, img->GetRequestedRegion());
it.Begin();
while( ! it.IsAtEnd() )
{
temp_vector = it.Get();
for (i = 0; i, 3>
*img)
{
typedef itk::Vector VectorType;
const unsigned int N = 3;
unsigned int i;
itk::ImageRegionIterator
it (img, img->GetRequestedRegion());
it.Begin();
while( ! it.IsAtEnd() )
{
for (i = 0; i ScalarImageType;
typedef itk::Image, 3> VectorImageType;
clock_t start, stop, no_adaptor_comp, adaptor_comp;
// Set up some images
itk::ImageRegion<3> region;
itk::Size<3> size;
size[0] = 100;
size[1] = 100;
size[2] = 100;
itk::Index<3> index;
index[0] = 0;
index[1] = 0;
index[2] = 0;
region.SetSize(size);
region.SetIndex(index);
ScalarImageType::Pointer scalar_image = ScalarImageType::New();
VectorImageType::Pointer vector_image = VectorImageType::New();
scalar_image->SetLargestPossibleRegion(region);
scalar_image->SetBufferedRegion(region);
scalar_image->SetRequestedRegion(region);
scalar_image->Allocate();
vector_image->SetLargestPossibleRegion(region);
vector_image->SetBufferedRegion(region);
vector_image->SetRequestedRegion(region);
vector_image->Allocate();
// Time trials
std::cout << "Speed of adaptor supporting interator (for reference) \t";
start = clock();
AdaptorSupportedIteratorSpeed(scalar_image);
stop = clock();
adaptor_comp = stop - start;
std::cout << adaptor_comp << std::endl;
std::cout << "Speed of interator that does not support adaptors (for
reference) \t";
start = clock();
NoAdaptorSupportIteratorSpeed(scalar_image);
stop = clock();
no_adaptor_comp = stop - start;
std::cout << no_adaptor_comp << std::endl;
std::cout << "Modifying scalar image using adaptor api...\t";
start = clock();
AdaptorSupportedModifyScalars(scalar_image);
stop = clock();
std::cout << (stop - start) << "\t compensated = " << (stop-start) -
adaptor_comp << std::endl;
std::cout << "Modifying scalar image bypassing adaptor api...\t";
start = clock();
NoAdaptorSupportModifyScalars(scalar_image);
stop = clock();
std::cout << (stop - start) << "\t compensated = " << (stop-start) -
no_adaptor_comp << std::endl;
std::cout << "Modifying vector image using adaptor api...\t";
start = clock();
AdaptorSupportedModifyVectors(vector_image);
stop = clock();
std::cout << (stop - start) << "\t compensated = " << (stop-start) -
adaptor_comp << std::endl;
std::cout << "Modifying vector image bypassing adaptor api...\t";
start = clock();
NoAdaptorSupportModifyVectors(vector_image);
stop = clock();
std::cout << (stop - start) << "\t compensated = " << (stop-start) -
no_adaptor_comp <what should I do if I want to attend the tcon? just dial in that number and
>code? should I sign-up for something in advance?
>BTW, please confirm that there is one tomorrow, since some of my group may
>sign in and discuss our current project with the Architecture persons.
>thanks,
>yinpeng
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From yj76@columbia.edu Fri, 16 Mar 2001 10:46:47 -0500
Date: Fri, 16 Mar 2001 10:46:47 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] Depth of nested templates
Hi, brad:
I just checked in another classes:
itkVoronoiSegmentationImageFilter
please try a compiler on you machine and told me if there is any
too-deep-templates problem. my compiler won't give any warning on
this.(MSVC)
thanks,
yinpeng
From ibanez@cs.unc.edu Fri, 16 Mar 2001 11:13:21 -0500
Date: Fri, 16 Mar 2001 11:13:21 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Image/ImageAdaptor/DataAccessor modification
Hi,
The questions raised by Josh's tests are quite relevant.
Let's consider them in order:
1) The difference in execution time between:
a- ImageIterator (the one not supporting adaptors) and
b- ImageIteratorWithIndex (the one supporting adaptors)
when no data is accessed, are due to the computation of the index.
(a) is faster because it doesn't keep an updated value of the index,
that's good because it doesn't penalize the user for something he
is not using. Should the user require to access the index, then (b)
will be faster because the index is updated progressively.
So far, this differences are not related to the fact of supporting
adaptors or not, as Josh correctly noted. The time corrections that
Josh did are a perfect compensation for this difference in nature
between the two iterators. So, the time comparisions after
compensation are fair.
2) The differences found in gcc in Linux for access to scalars are
surprisingly high. As far as I can see, they seem to be caused
by the fact that the Get() method in the default DataAccessor class
is returning by value (copy) instead of by reference:
TExternalType Get( const TInternalType & value ) const {
return (TExternalType) value; }
So, I just modified to return by const reference like:
const TExternalType & Get( const TInternalType & value ) const {
return (TExternalType) value; }
This is posible only in the default DataAccessor because the
InternalType and the ExternalType are the same. In any other real
DataAccessor, those types will be different, and a temporary copy
of the data will be required, in those cases the return has to be
done by copy; but that's fair because the user will be paying for
the 'cast' which is somethig he is really using.
Curiously SGI do better, and seems to optimize the inlined copy...
3) The vector case is the real difficult point. In fact what make
them special is not that they are vectors, but that they are big
objects and the algorithm needs to modify them 'in place'. For this
case, the Get()/Set() approach is not efficient because it forces
the user to make two additional copies of the object data.
It seems that the best solution for this case is exactly what Josh
proprosed: "Bypass the data accessor".
It is hard to argue otherwise given that this is in the heart of
access to image data, and will affect the overall performance of
the toolkit.
The option is then, to provide an additional method for accessing
pixel data. The use of this method will be justified when data
should be modified in place, like the += vector operation in Josh's
example. In any other cases, when data is only read or only writen,
there is not real penalty in using the Get()/Set() methods.
The additional access method could be the * operator, like was
originaly, of something like the Value() method in the VectorContainer
iterators.
*it += vectorInc; or it.Value() += vectorInc;
The * operator seems to be a more natural choice, which could be bad,
because programmers will have more tendency to use it even if the can
use Set()/Get() and support adaptors. The Value() method has the
advantage of giving a uniform aspect to all the interators on the
toolkit, and that it will force programmers to think if they really
need Value() or can go with Set()/Get().
What are your preferences ?
Note that supporting Adaptors is specially important in the input
and output of filters, but any other computation performed on
images declared internally can count on the real nature of the
images (unless the type of the input or output has been used to
declare them, a case which also needs further discussion...).
Thanks,
Luis
PS.
Unfortunately I'll not be able to make the TCon today.
I'll appreciate if somebody can summarize the feedback
from other people and the discussion on this topic
today, and post it to the list. Sorry about that.
From brad.king@kitware.com Fri, 16 Mar 2001 11:35:56 -0500 (EST)
Date: Fri, 16 Mar 2001 11:35:56 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Depth of nested templates
Yipeng,
> I just checked in another classes: itkVoronoiSegmentationImageFilter
> please try a compiler on you machine and told me if there is any
> too-deep-templates problem. my compiler won't give any warning on
> this.(MSVC)
After a couple minor repairs, it works fine on GCC (see the log entry
when you update). There are still some signed/unsigned warnings for you
to hammer out, though.
If something builds correctly on your development environment, and you are
confident it is correct C++ code, go ahead and check it in. One of the
dashboard's purposes is to let you see how your changes affected other
platforms. You can (should) check it the next day to see if you broke
anything.
-Brad
From yj76@columbia.edu Fri, 16 Mar 2001 12:58:23 -0500
Date: Fri, 16 Mar 2001 12:58:23 -0500
From: Yinpeng Jin yj76@columbia.edu
Subject: [Insight-developers] TCON
Hi, all:
Since we(Upenn & Columbia Segmentation group) had checked in some classes,
and we would like to have some feedbacks from the architecture point of
view. So if you have any question/suggestion on our codes, please join us at
2:00pm today for the TCON.
Thanks.
Yinpeng
From hughett@mercur.uphs.upenn.edu Fri, 16 Mar 2001 14:57:25 -0500
Date: Fri, 16 Mar 2001 14:57:25 -0500
From: Paul Hughett hughett@mercur.uphs.upenn.edu
Subject: [Insight-developers] itkImageMomentsTest on LinuxIA64 and on WinNT/VC++
The dashboard reports that itkImageMomentsTest fails to run on the
Linux IA64 platform, which--as I understand it--means that the test
program failed to compile. Alas, I have no IA64 platform to compile
on. Could someone who does compile itkImageMomentsTest for me and
send me the compiler messages?
Similarly, the same program compiles on WinNT/VC++ but has very large
errors, while running fine on the platform I have to test it on here.
Could someone compile it on that platform and send me the compiler
messages? Perhaps one of the warnings will shed some light as to
why it breaks so badly on that one platform.
Many thanks.
Paul Hughett
PS That's in the Testing/Code/Algorithms directory. pH
From wlorens1@nycap.rr.com Fri, 16 Mar 2001 20:06:41 -0500
Date: Fri, 16 Mar 2001 20:06:41 -0500
From: Bill Lorensen wlorens1@nycap.rr.com
Subject: [Insight-developers] itkImageMomentsTest on LinuxIA64 and
on WinNT/VC++
We just added that platform. It is still not building correctly. Not your problem. Yet...
At 02:57 PM 3/16/01 -0500, Paul Hughett wrote:
>The dashboard reports that itkImageMomentsTest fails to run on the
>Linux IA64 platform, which--as I understand it--means that the test
>program failed to compile. Alas, I have no IA64 platform to compile
>on. Could someone who does compile itkImageMomentsTest for me and
>send me the compiler messages?
>
>Similarly, the same program compiles on WinNT/VC++ but has very large
>errors, while running fine on the platform I have to test it on here.
>Could someone compile it on that platform and send me the compiler
>messages? Perhaps one of the warnings will shed some light as to
>why it breaks so badly on that one platform.
>
>Many thanks.
>
>
>Paul Hughett
>
>PS That's in the Testing/Code/Algorithms directory. pH
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From ibanez@cs.unc.edu Sun, 18 Mar 2001 13:36:03 -0500
Date: Sun, 18 Mar 2001 13:36:03 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkMatrix and the binary operator *
Hi,
There is an error in itkMatrix, that I'm trying to fix.
It is related with the operator* with another matrix.
itkMatrix derives from vnl_matrix_fixed, which
derives from vnl_matrix.
The operator* is defined for vnl_matrix, but
it doesn't seems to recognize itkMatrix as a
valid argument.
The strange thing is that it seems to work only
when itkMatrix is declared of type double.
>From the dashboard, it seems that it only happens
in VC++. I've seen that the definition of the *
operator in vnl_matrix is done differently in vnl
depending on the platform. (using ifdefs)
Any ideas ?
Thanks
Luis
From brad.king@kitware.com Tue, 20 Mar 2001 16:26:53 -0500 (EST)
Date: Tue, 20 Mar 2001 16:26:53 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] CMake/Cable (EVERYONE PLEASE READ)
Hello, all:
One of the problems we have all been having with testing our code is
that not all the code in a template class is necessarily compiled by
our test programs. Consider the following example:
--------------------------------------------------
template
class MyVector
{
public:
typedef TElement ElementType;
void MyMethod0() {}
void MyMethod1() { callingNonexistentFunction(); }
};
int main()
{
MyVector mv; // Implicit instantiation of MyVector
mv.MyMethod0(); // Only the methods that are called will
// be compiled.
return 0;
}
--------------------------------------------------
This program will compile without error because MyMethod1() is never
instantiated, and therefore its code is not compiled. However, if
someone else uses MyVector, and tries to call MyMethod1(), they will
get a compiler error even though their own code is perfectly valid.
If, however, there were an explicit instantiation of MyVector,
then MyMethod1() would be compiled, even if it is not called. An
explicit instantiation would look like "template class MyVector;"
and would produce a compiler error immediately on MyMethod1().
Eventually, we would also like to be able to build explicit
instantiations of ITK classes into the libraries to reduce code bloat.
Writing out dozens of combinations of template arguments to create
these instantiations is time consuming and hard to maintain. The new
tool "CABLE" I have added to Insight solves this problem, and an
extension to CMake makes it easy to use as part of the build process.
Near the end of last week I checked in changes to Code/CMakeLists.txt,
Code/Common/CMakeLists.txt, and Code/BasicFilters/CMakeLists.txt that
introduce automatic generation of explicit instantiations of template
classes into the Common and BasicFilters libraries. Now that I have
confirmed that the builds with the instantiations are working on all
of the dashboard's platforms, it is time to announce this so that
everyone can use the new CMake / CABLE features.
--------------------------------------------------
Here is a short tutorial of how to use the CMake-Cable-Command
features. First off, all control is done in the CMakeLists.txt files
through CMake commands. All Cable commands in CMake begin with
"CABLE_" to distinguish them from the rest of CMake's commands.
The idea here is to define sets of types that can be used as template
arguments and then use these sets to automatically instantiate various
combinations of template classes.
A set is defined with the CABLE_DEFINE_SET command. For example
CABLE_DEFINE_SET (BasicType int float long "unsigned short")
defines a set called "BasicType" with members "int", "float", "long",
and "unsigned short". This set can be referenced in certain other
CMake CABLE_ commands by its name with a '$' in front. Now
we define a set of specializations of the above "MyVector" class
with each of these basic types.
CABLE_DEFINE_SET (MyVectorType "MyVector<$BasicType>")
This defines a set called "MyVectorType" with the following members:
MyVector
MyVector
MyVector
MyVector
Now, say we want to actually tell CMake to put explicit instantiations
of MyVector into the library being built. Assume that we have fixed
the code in MyMethod1() so that it will compile. Here, I assume that
we defined "MyVector" in the files myVector.h and myVector.txx CMake
will automatically find the file extensions when the source is
specified below. Now, we add this code to CMakeLists.txt:
# Define the name of the package in which we wish to put the
# instantiations. This controls the file name with the generated
# code, and must come before any CABLE_INSTANTIATE_CLASS commands.
CABLE_PACKAGE(MyVectors)
# We must tell CMake where to find the source to our template class,
# since it cannot be listed in the regular SOURCE_FILES command.
CABLE_SOURCE_FILES(myVector)
# Create explicit instantiations of the class.
CABLE_INSTANTIATE_CLASS($MyVectorType
"MyVector")
This will create a package called "MyVectors" with the following
explicit instantiations:
MyVector
MyVector
MyVector
MyVector
MyVector
This package will automatcially be compiled and linked into the
library specified in this CMakeLists file.
The above tutorial is just a quick overview of some of the most
basic features in Cable and the CMake extension controlling it.
Code already checked in to the Code/CMakeLists.txt,
Code/Common/CMakeLists.txt, and Code/BasicFilters/CMakeLists.txt
demonstrates much more of the functionality. To help you read it,
here is a quick description of some of the features it uses:
In a CABLE_DEFINE_SET command, the keyword "SOURCE_FILES" can be used
after the list of set elements. Any argument after this keyword will
be interpreted as a source file specification. Any place that
references the set will automatically include the sources specified.
When referencing a set with a '$', the set name is terminated when
a character other than A-Z, a-z, 0-9, '_', or ':' is encountered.
This prevents one from using the members of a set to construct
an identifier. Alternatively, $(setname) will use all the characters
between the parentheses as the set name. This allows one to write
something like:
CABLE_DEFINE_SET(digits 0 1 2 3 4 5 6 7 8 9)
CABLE_DEFINE_SET(zeroTo99 "$(digits)$digits")
The set "zeroTo99" will contain "00" "01" "02" ... "97" "98" "99".
--------------------------------------------------
I would like everyone to read the example code already provided in the
CMakeLists files like Code/Common/CMakeLists.txt and add code
to create one instantiation of each of your own template classes.
Feel free to email the list with questions, and I will answer them.
The commands are very flexible and may take some "getting-used-to"
to use them.
We would like to get an explicit instantiation of every template class
in all of Insight so that we are sure all code is compiled for the
dashboard. These instantiations will be linked into the libraries
where they are defined, and can be used in any test program. For now,
however, we would like to stick to instantiations using the
ScalarType, PixelType, Dimension, PointType, VectorType, ImageType,
and MeshType sets already defined in Code/CMakeLists.txt. This means
that filters should only be instantiated as
"MyFilter<$ImageType,$ImageType>" (assuming that MyFilter is templated
over TInputImage, TOutputImage). However, feel free to define other
sets to shorten your code (look at the "iaFunc" example already
present).
Good luck, and thank you for taking the time to read this,
-Brad
P.S. In case anyone is wondering what "CABLE" means, it is a
recursive acronym:
CABLE Automates Bindings for Language Extension
Eventually this will become the Tcl/Python/etc wrapping tool.
From zhuge@mipg.upenn.edu Tue, 20 Mar 2001 17:45:57 -0600
Date: Tue, 20 Mar 2001 17:45:57 -0600
From: Ying Zhuge zhuge@mipg.upenn.edu
Subject: [Insight-developers] compile error
I just update the Insight. Something wrong with the ITKCommon.lib and
ITKBasicFilters.lib when compiling
fatal error C1083: Cannot open source file:
'C:\Users\Insight-VC++\Code\BasicFilters\Cxx\BasicFilters_cxx.cxx': No
such file or directory
fatal error C1083: Cannot open source file:
'C:\Users\Insight-VC++\Code\Common\Cxx\Common_cxx.cxx': No such file or
directory
Please check. Thanks a lot
Ying
From brad.king@kitware.com Tue, 20 Mar 2001 18:16:14 -0500 (EST)
Date: Tue, 20 Mar 2001 18:16:14 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] txx policy
Hello, all:
Since we are beginning to use explicit instantiations, the
ITK_MANUAL_INSTANTIATION flag may not always be left undefined now. This
means that .txx files must be #included when instantiations are to be
created. To prevent multiple inclusion, we must begin a new policy for
.txx files similar to that of .h files. All txx files must be surrounded
by the following code:
#ifndef _NameOfFileWithoutExtension_txx
#define _NameOfFileWithoutExtension_txx
// rest of code in .txx source.
#endif
I used an emacs script to put this code around every txx currently checked
in (other than those in vxl) and commited the changes. However, anyone
writing a new .txx source must add this code, just as when writing a .h.
-Brad
From brad.king@kitware.com Tue, 20 Mar 2001 18:22:12 -0500 (EST)
Date: Tue, 20 Mar 2001 18:22:12 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] compile error
> I just update the Insight. Something wrong with the ITKCommon.lib and
> ITKBasicFilters.lib when compiling
>
> fatal error C1083: Cannot open source file:
> 'C:\Users\Insight-VC++\Code\BasicFilters\Cxx\BasicFilters_cxx.cxx': No
> such file or directory
>
> fatal error C1083: Cannot open source file:
> 'C:\Users\Insight-VC++\Code\Common\Cxx\Common_cxx.cxx': No such file or
> directory
Did you re-run CMakeSetup?
-Brad
From george@stetten.com Tue, 20 Mar 2001 19:40:59 -0500
Date: Tue, 20 Mar 2001 19:40:59 -0500
From: George Stetten george@stetten.com
Subject: [Insight-developers] what to compile
I assume we all don't really need to be compiling all the test programs.
What's an easy way to avoid compiling them that can be easily explained to any
newcomer?
George Stetten
From brad.king@kitware.com Wed, 21 Mar 2001 08:59:57 -0500 (EST)
Date: Wed, 21 Mar 2001 08:59:57 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] what to compile
George,
> I assume we all don't really need to be compiling all the test
> programs. What's an easy way to avoid compiling them that can be
> easily explained to any newcomer?
You should be able to just remove the "Testing" and "Examples"
subdirectories from the "SUBDIRS" command at the top level CMakeLists.txt
file. If you are doing development, though, you should be compiling the
tests to make sure you didn't break anything with your changes before you
check them in. Also, until we get this explicit-instantiation stuff in
the libraries in full use, the test cause most of the build to occur, and
almost nothing will build without them.
A better way might be to remove indiviudal tests by commenting them
out. You can go to the "TESTS" command in the CMakeLists.txt of
Testing/Code/* directories and put a "#" before any test, and it will be
ignored.
-Brad
From chandra@cs.unc.edu Thu, 22 Mar 2001 01:16:43 -0500
Date: Thu, 22 Mar 2001 01:16:43 -0500
From: Parag Chandra chandra@cs.unc.edu
Subject: [Insight-developers] Changes to ImageIO
This is a multi-part message in MIME format.
------=_NextPart_000_06CE_01C0B26D.C16F9A70
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Hi,
If any of you have written file readers that derive from ImageIO, please =
make note of the following change. In order to accomodate filetypes that =
use multiple extensions (for example, jpeg sometimes uses .jpeg in =
addition to .jpg), I modified ImageIO::GetSupportedFileExtensions() =
from:
virtual std::string GetSupportedFileExtensions() const =3D 0;
to:
typedef std::deque FileExtensionsListType;
virtual FileExtensionsListType GetSupportedFileExtensions() const =3D 0;
So now, instead of returning a single string, it returns a list of =
strings. The changes that need to be made to any readers are shown in =
Unsupported/MetaImage/FileIOMetaImage.cxx and .h. Besides the one for =
MetaImage, I'm not aware of any other readers, but if there are, please =
let me know and I will help you make the necessary changes.
Thanks,
-Parag
------=_NextPart_000_06CE_01C0B26D.C16F9A70
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Hi,
If any of you have written file readers =
that derive=20
from ImageIO, please make note of the following change. In order to =
accomodate=20
filetypes that use multiple extensions (for example, jpeg sometimes uses =
.jpeg=20
in addition to .jpg), I modified ImageIO::GetSupportedFileExtensions()=20
from:
virtual std::string =
GetSupportedFileExtensions()=20
const =3D 0;
to:
typedef std::deque<std::string>=20
FileExtensionsListType;
virtual FileExtensionsListType=20
GetSupportedFileExtensions() const =3D 0;
So now, instead of returning a single =
string, it=20
returns a list of strings. The changes that need to be made to any =
readers are=20
shown in Unsupported/MetaImage/FileIOMetaImage.cxx and .h. Besides the =
one for=20
MetaImage, I'm not aware of any other readers, but if there are, please =
let me=20
know and I will help you make the necessary changes.
Thanks,
-Parag
------=_NextPart_000_06CE_01C0B26D.C16F9A70--
From ibanez@cs.unc.edu Thu, 22 Mar 2001 10:11:38 -0500
Date: Thu, 22 Mar 2001 10:11:38 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Iterators bypassing DataAccessor / ImageAdaptors
Hi Josh,
I just added a Value() method to ImageIteratorWithIndex and to
ImageConstIteratorWithIndex.
This method bypasses the DataAccessor, and return a direct
reference to the pixel data. That should provide a faster access
to pixel data, at the price of not supporting ImageAdaptors.
The use will be something like:
it.Begin();
while( it.IsAtEnd() ) {
it.Value() = 5;
a = it.Value();
it.Value() += 5;
++it;
}
There were also a couple of changes in the way iterators access
data, and that may improve the performance when using Set/Get.
Could you check in your code for testing iterator speed,
maybe as a test program ?
Thanks a lot
Luis
From cates@cayenne.cs.utah.edu Thu, 22 Mar 2001 10:21:42 -0700 (MST)
Date: Thu, 22 Mar 2001 10:21:42 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] Iterators bypassing DataAccessor /
ImageAdaptors
Hi Luis,
OK. I'll check in my test under the Testing/../Common directory today.
I'll add a few more tests that use the new API, too.
There was some adaptor discussion at last Friday's TCON about how to
support both modes of access while minimizing complexity for developers
and users. I don't believe any major conclusions were reached except that
the issue needs more study. If you have some specific examples of how you
are using image adaptors with your algorithms, these would be good as a
reference for discussion.
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
On Thu, 22 Mar 2001, Luis Ibanez wrote:
> Hi Josh,
>
> I just added a Value() method to ImageIteratorWithIndex and to
> ImageConstIteratorWithIndex.
>
> This method bypasses the DataAccessor, and return a direct
> reference to the pixel data. That should provide a faster access
> to pixel data, at the price of not supporting ImageAdaptors.
>
> The use will be something like:
>
> it.Begin();
> while( it.IsAtEnd() ) {
> it.Value() = 5;
> a = it.Value();
> it.Value() += 5;
> ++it;
> }
>
>
> There were also a couple of changes in the way iterators access
> data, and that may improve the performance when using Set/Get.
>
> Could you check in your code for testing iterator speed,
> maybe as a test program ?
>
>
> Thanks a lot
>
>
>
> Luis
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From george@stetten.com Thu, 22 Mar 2001 11:32:13 -0500
Date: Thu, 22 Mar 2001 11:32:13 -0500
From: George Stetten george@stetten.com
Subject: [Insight-developers] Iterators bypassing DataAccessor /ImageAdaptors
Is there going to be official documentation for the alpha release? An
overview would be very handy, and we could all be looking it over.
George Stetten
From will.schroeder@kitware.com Thu, 22 Mar 2001 14:15:00 -0500
Date: Thu, 22 Mar 2001 14:15:00 -0500
From: Will Schroeder will.schroeder@kitware.com
Subject: [Insight-developers] PointSet checked in
Hi Folks-
I checked in the PointSet class. It's a superclass of Mesh. I've added a test (Testing/Code/Common/itkPointSetTest.cxx) and added it to CABLE's instantiation classes. I've got some clean up to do...but please let me know if there are any problems.
Will
From aylward@unc.edu Thu, 22 Mar 2001 14:13:00 -0500
Date: Thu, 22 Mar 2001 14:13:00 -0500
From: Stephen R. Aylward aylward@unc.edu
Subject: [Insight-developers] PointSet checked in
This is great news! Thanks!
We'll switch our landmark stuff to use it!
Stephen
Will Schroeder wrote:
>
> Hi Folks-
>
> I checked in the PointSet class. It's a superclass of Mesh. I've added a test (Testing/Code/Common/itkPointSetTest.cxx) and added it to CABLE's instantiation classes. I've got some clean up to do...but please let me know if there are any problems.
>
> Will
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
--
===============================================
Stephen R. Aylward
Assistant Professor of Radiology
Adjunct Assistant Professor of Computer Science
http://www.cs.unc.edu/~aylward
aylward@unc.edu
(919) 966-9695
From ibanez@cs.unc.edu Thu, 22 Mar 2001 17:26:58 -0500
Date: Thu, 22 Mar 2001 17:26:58 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Disk full
Hi,
I got this strange error message compiling the toolkit,
and found that the binaries are taken a considerable
amount of space.
I know that 40Gb disks are getting cheaper and cheaper...
but I wonder if I'm doing something wrong in my local build.
Using Cygwin, and making an outsource build,
the binaries took :
439 Mb (well,... the disk was full at this point).
These are the sizes of some of the libraries:
BasicLibraries 1.5 Mb
Common 9.6 Mb
Numerics 24.7 Mb
The Testing directory took: 331 Mb !
Algorithms 91 Mb
BasicFilters 124 Mb
Common 64 Mb
Numerics 48 Mb
------
Under WinNT, the binaries are more affordable,
the take only: 144 Mb.
(That should be wrong... VC++ is doing better that gcc !)
Is there something that can be done to reduce the size ?
Is that consequence of the manual instantiation ?
Thanks
Luis
From chandra@cs.unc.edu Thu, 22 Mar 2001 17:32:41 -0500
Date: Thu, 22 Mar 2001 17:32:41 -0500
From: Parag Chandra chandra@cs.unc.edu
Subject: [Insight-developers] Disk full
Luis, I've noticed the same thing. In my case though, I think VC++ is taking
up less space than Cygwin because with VC++ it's much easier to just
selectively compile the projects you need, and that's what I have done. With
Cygwin on the other hand, I just type make from the root directory and it
builds everything, because I'm lazy ;-) I don't know if Cygwin works the
same way, but if I compile everything in VC++ under the Release target, then
all the debug information is left out and the compiled code is much smaller.
Perhaps there's some way to do this under Cygwin as well?
-Parag
From aylward@unc.edu Thu, 22 Mar 2001 18:01:58 -0500
Date: Thu, 22 Mar 2001 18:01:58 -0500
From: Stephen R. Aylward aylward@unc.edu
Subject: [Insight-developers] [Fwd: Changes to MetaImage...]
How should we reference data files in the example directories if the
build is placing the executables in different directories? See below
for more details....
Thanks,
Stephen
Parag Chandra wrote:
>
> Ok, that solves half the problem. I'll put the test and the accompanying
> files into examples. But how would I reference that (the examples) directory from within
> the test program? Each platform will place the executable for this test into
> something like ${CMAKE_BINARY_DIR}/Examples, but ${CMAKE_BINARY_DIR} isn't
> even an environment variable for me to reference, it's a variable that's
> internal to CMake. If I reference the file without the pathname, like
> "TestInput.mhd", then the file has to be in the same directory as the
> executable, and that means I would have to place a copy of the test file
> into Insight-Cygwin, Insight-VC++, Insight-Linux, etc. Not only is that
> redundant, but there's no way I could foresee the additional platforms we
> might target, and so the test running on those platforms will generate an
> exception when it can't find the file. The only thing I can see is to place
> the test file into one directory, and then run the test for each platform
> from that directory by using the full pathname of the executable, e.g.
> /Insight-Cygwin/Examples/itkFileIOMetaImageTest.exe. Does anyone have any
> other ideas?
>
> Thanks,
> -Parag
>
> ---
> You are currently subscribed to caddlab as: aylward@unc.edu
> To unsubscribe send a blank email to leave-caddlab-706N@listserv.unc.edu
--
===============================================
Stephen R. Aylward
Assistant Professor of Radiology
Adjunct Assistant Professor of Computer Science
http://www.cs.unc.edu/~aylward
aylward@unc.edu
(919) 966-9695
From brad.king@kitware.com Thu, 22 Mar 2001 18:23:03 -0500 (EST)
Date: Thu, 22 Mar 2001 18:23:03 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Disk full
Luis,
> Is there something that can be done to reduce the size ?
> Is that consequence of the manual instantiation ?
Actually, it is because we are NOT doing manual instantiation for the
tests. Every template class that is needed by each test is being
instantiated and linked statically into the executable for that test.
This means we have a copy of many template classes in every test
exectuable. The idea of the manual instantiation is to be able to put the
instantiations you need in the libraries, and link to them from the
executables. This way, each instantiation only exists in one place, and
the code will be much smaller.
-Brad
From chenting@graphics.cis.upenn.edu Thu, 22 Mar 2001 18:11:29 -0600
Date: Thu, 22 Mar 2001 18:11:29 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] about dataobject update
Hi! I am working on developing a recursive segmentation method. But I met
this problem. As for itkMRFImageFilter, when I make change to the training
image and want to update the segmentation result. The filter seems just did
not respond to the update of training image. The problem is at:
itkDataObject.cxx:
void
DataObject
::UpdateOutputData()
{
// If we need to update due to PipelineMTime, or the fact that our
// data was released, then propagate the UpdateOutputData to the source
// if there is one.
>>>>>>>cannot enter the outer if{...}
if ( m_UpdateTime < m_PipelineMTime || m_DataReleased ||
this->RequestedRegionIsOutsideOfTheBufferedRegion())
{
if ( m_Source )
{
m_Source->UpdateOutputData(this);
}
}
}
it seems I have to do something to make it know it is time for update, but
where and what should do? Any ideas?
ting
From wlorens1@nycap.rr.com Thu, 22 Mar 2001 19:54:38 -0500
Date: Thu, 22 Mar 2001 19:54:38 -0500
From: Bill Lorensen wlorens1@nycap.rr.com
Subject: [Insight-developers] Disk full
I build shared libraries. Use the --with-shared on configure.
I'll check my sizes when I get to work tomorrow.
Bill
At 05:26 PM 3/22/01 -0500, Luis Ibanez wrote:
>Hi,
>
>I got this strange error message compiling the toolkit,
>and found that the binaries are taken a considerable
>amount of space.
>
>I know that 40Gb disks are getting cheaper and cheaper...
>but I wonder if I'm doing something wrong in my local build.
>
>Using Cygwin, and making an outsource build,
>the binaries took :
>
> 439 Mb (well,... the disk was full at this point).
>
> These are the sizes of some of the libraries:
>
> BasicLibraries 1.5 Mb
> Common 9.6 Mb
> Numerics 24.7 Mb
>
> The Testing directory took: 331 Mb !
>
> Algorithms 91 Mb
> BasicFilters 124 Mb
> Common 64 Mb
> Numerics 48 Mb
>
>
>------
>
>Under WinNT, the binaries are more affordable,
>the take only: 144 Mb.
>(That should be wrong... VC++ is doing better that gcc !)
>
>
>Is there something that can be done to reduce the size ?
>
>Is that consequence of the manual instantiation ?
>
>
>
>
>Thanks
>
>
>Luis
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From bill.hoffman@kitware.com Thu, 22 Mar 2001 22:39:01 -0500
Date: Thu, 22 Mar 2001 22:39:01 -0500
From: Bill Hoffman bill.hoffman@kitware.com
Subject: [Insight-developers] Disk full
Configure picks -O2 -g for gcc, which is REALLY HUGE for cygwin.
Just -g or -O2 should be quite a bit smaller. Visual C++ uses
those program database files to keep the debug info central, and
gcc duplicates a lot of the debug info into each .o file, and of course
it does not truncate symbols longer than 255, as we know VC++ does.
-Bill (father of Max) Hoffman
http://www.setonhealth.org/html/ChildB2.html (see Maxwell Robert)
At 07:54 PM 3/22/01 -0500, Bill Lorensen wrote:
>I build shared libraries. Use the --with-shared on configure.
>
>I'll check my sizes when I get to work tomorrow.
>
>Bill
>
>At 05:26 PM 3/22/01 -0500, Luis Ibanez wrote:
>
> >Hi,
> >
> >I got this strange error message compiling the toolkit,
> >and found that the binaries are taken a considerable
> >amount of space.
> >
> >I know that 40Gb disks are getting cheaper and cheaper...
> >but I wonder if I'm doing something wrong in my local build.
> >
> >Using Cygwin, and making an outsource build,
> >the binaries took :
> >
> > 439 Mb (well,... the disk was full at this point).
> >
> > These are the sizes of some of the libraries:
> >
> > BasicLibraries 1.5 Mb
> > Common 9.6 Mb
> > Numerics 24.7 Mb
> >
> > The Testing directory took: 331 Mb !
> >
> > Algorithms 91 Mb
> > BasicFilters 124 Mb
> > Common 64 Mb
> > Numerics 48 Mb
> >
> >
> >------
> >
> >Under WinNT, the binaries are more affordable,
> >the take only: 144 Mb.
> >(That should be wrong... VC++ is doing better that gcc !)
> >
> >
> >Is there something that can be done to reduce the size ?
> >
> >Is that consequence of the manual instantiation ?
> >
> >
> >
> >
> >Thanks
> >
> >
> >Luis
> >
> >_______________________________________________
> >Insight-developers mailing list
> >Insight-developers@public.kitware.com
> >http://public.kitware.com/mailman/listinfo/insight-developers
>
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From millerjv@crd.ge.com Fri, 23 Mar 2001 07:37:49 -0500
Date: Fri, 23 Mar 2001 07:37:49 -0500
From: Miller, James V (CRD) millerjv@crd.ge.com
Subject: [Insight-developers] about dataobject update
A change to a pixel does not cause the DataObject to think it has been modified. There are two
reasons for this. First, if we ticked the modified time for the data object for every pixel
modification, we would quickly roll over the modified time ivar. Second, we'd hate to have to tick
the modified time for every pixel modification (that would be an extra add per pixel change).
Once you have finished making all your changes to the pixel data, send the DataObject a Modified()
call. This will mark the data object as being modified and your pipeline should re-execute.
Jim
-----Original Message-----
From: Ting Chen [mailto:chenting@graphics.cis.upenn.edu]
Sent: Thursday, March 22, 2001 7:11 PM
To: Insight Developers
Subject: [Insight-developers] about dataobject update
Hi! I am working on developing a recursive segmentation method. But I met
this problem. As for itkMRFImageFilter, when I make change to the training
image and want to update the segmentation result. The filter seems just did
not respond to the update of training image. The problem is at:
itkDataObject.cxx:
void
DataObject
::UpdateOutputData()
{
// If we need to update due to PipelineMTime, or the fact that our
// data was released, then propagate the UpdateOutputData to the source
// if there is one.
>>>>>>>cannot enter the outer if{...}
if ( m_UpdateTime < m_PipelineMTime || m_DataReleased ||
this->RequestedRegionIsOutsideOfTheBufferedRegion())
{
if ( m_Source )
{
m_Source->UpdateOutputData(this);
}
}
}
it seems I have to do something to make it know it is time for update, but
where and what should do? Any ideas?
ting
_______________________________________________
Insight-developers mailing list
Insight-developers@public.kitware.com
http://public.kitware.com/mailman/listinfo/insight-developers
From blezek@crd.ge.com Fri, 23 Mar 2001 08:24:27 -0500 (EST)
Date: Fri, 23 Mar 2001 08:24:27 -0500 (EST)
From: Daniel J. Blezek, Ph.D. blezek@crd.ge.com
Subject: [Insight-developers] [Fwd: Changes to MetaImage...]
Stephen and Parag,
This exact issue drove the discussion about file reading and writing at
last week's telephone conference. I have several ideas in mind as to
finding images and writing images for regression testing, but have not
made them happen just yet. In the mean time, I would welcome any
suggestions you may have as to how to address this issue. In our
experience with VTK, which may be different for ITK, some platforms
require a platform specific "gold image", and require special processing.
Here are some of my ideas, strictly related to regression testing:
- Have a Toolkit similar to Java, so to get the name of a file you would
do something like this:
std::string s = Toolkit::GetFileName ( "VHF/Head.png" );
The Toolkit would properly expand the file name to be the path to the
image you want, allowing images to be stored on fixed media(CDROM). On
the outbound side, you would request an output image in a similar manner:
std::string s = Toolkit::GetImageTestResultFileName (
"itkFileIOMetaImageTest" );
The Toolkit would expand this name to a directory where test images
sit, and a latter process would pick up the images and compare them to the
"gold standard". If the test were run in an example mode, it would simply
write the output in the current directory, for the user to take a look at
later.
- This raises the issue of file formats yet again. Everyone seems to cook
up their own file format to meet the needs of whatever problem they have
at hand! (We are no exception...) In my view, the current problem at hand
is regression testing. I've given this issue some thought. We need:
- a file format that is easily displayed locally and through a browser
- a format that is easily created by all Insight developers
- a format that has 2D support at minimum
- a format that supports as many common datatypes as possible
The first point drives the selection down considerably. The only real
canidates are Gif, Jpeg, and PNG. Gif and Jpeg are generally lossy
compression, and don't fit the bill. PNG is a relatively new format, with
good support from most OS/viewers/browsers. PNG is also licensed for use
in a toolkit such as ITK. The real downside is that PNG only supports 8
and 16 bit datatypes. I feel this is acceptable because medical image
data generally comes in these two types, so that should not be a limiting
factor. PNG support is built and working on my local checkout.
Since I'm doing most of the testing infrastructure, I will offer to serve
as the focal point for this issue, and welcome any suggestions.
Thanks,
-dan
On Thu, 22 Mar 2001, Stephen R. Aylward wrote:
>
> How should we reference data files in the example directories if the
> build is placing the executables in different directories? See below
> for more details....
>
> Thanks,
> Stephen
>
> Parag Chandra wrote:
> >
> > Ok, that solves half the problem. I'll put the test and the accompanying
> > files into examples. But how would I reference that (the examples) directory from within
> > the test program? Each platform will place the executable for this test into
> > something like ${CMAKE_BINARY_DIR}/Examples, but ${CMAKE_BINARY_DIR} isn't
> > even an environment variable for me to reference, it's a variable that's
> > internal to CMake. If I reference the file without the pathname, like
> > "TestInput.mhd", then the file has to be in the same directory as the
> > executable, and that means I would have to place a copy of the test file
> > into Insight-Cygwin, Insight-VC++, Insight-Linux, etc. Not only is that
> > redundant, but there's no way I could foresee the additional platforms we
> > might target, and so the test running on those platforms will generate an
> > exception when it can't find the file. The only thing I can see is to place
> > the test file into one directory, and then run the test for each platform
> > from that directory by using the full pathname of the executable, e.g.
> > /Insight-Cygwin/Examples/itkFileIOMetaImageTest.exe. Does anyone have any
> > other ideas?
> >
> > Thanks,
> > -Parag
> >
> > ---
> > You are currently subscribed to caddlab as: aylward@unc.edu
> > To unsubscribe send a blank email to leave-caddlab-706N@listserv.unc.edu
>
> --
> ===============================================
> Stephen R. Aylward
> Assistant Professor of Radiology
> Adjunct Assistant Professor of Computer Science
> http://www.cs.unc.edu/~aylward
> aylward@unc.edu
> (919) 966-9695
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
--
Daniel Blezek, Ph.D.
blezek@crd.ge.com
Visualization and Computer Vision Program
Electronic Systems Lab
GE Corporate Research & Development
From aylward@unc.edu Fri, 23 Mar 2001 13:52:59 -0500
Date: Fri, 23 Mar 2001 13:52:59 -0500
From: Stephen R. Aylward aylward@unc.edu
Subject: [Insight-developers] [Fwd: Changes to MetaImage...]
The only difficulty is that PNG doesn't provide element spacing
information. How about a medical format - e.g., analyze?
Stephen
"Daniel J. Blezek, Ph.D." wrote:
>
> Stephen and Parag,
>
> This exact issue drove the discussion about file reading and writing at
> last week's telephone conference. I have several ideas in mind as to
> finding images and writing images for regression testing, but have not
> made them happen just yet. In the mean time, I would welcome any
> suggestions you may have as to how to address this issue. In our
> experience with VTK, which may be different for ITK, some platforms
> require a platform specific "gold image", and require special processing.
>
> Here are some of my ideas, strictly related to regression testing:
>
> - Have a Toolkit similar to Java, so to get the name of a file you would
> do something like this:
>
> std::string s = Toolkit::GetFileName ( "VHF/Head.png" );
>
> The Toolkit would properly expand the file name to be the path to the
> image you want, allowing images to be stored on fixed media(CDROM). On
> the outbound side, you would request an output image in a similar manner:
>
> std::string s = Toolkit::GetImageTestResultFileName (
> "itkFileIOMetaImageTest" );
>
> The Toolkit would expand this name to a directory where test images
> sit, and a latter process would pick up the images and compare them to the
> "gold standard". If the test were run in an example mode, it would simply
> write the output in the current directory, for the user to take a look at
> later.
>
> - This raises the issue of file formats yet again. Everyone seems to cook
> up their own file format to meet the needs of whatever problem they have
> at hand! (We are no exception...) In my view, the current problem at hand
> is regression testing. I've given this issue some thought. We need:
>
> - a file format that is easily displayed locally and through a browser
> - a format that is easily created by all Insight developers
> - a format that has 2D support at minimum
> - a format that supports as many common datatypes as possible
>
> The first point drives the selection down considerably. The only real
> canidates are Gif, Jpeg, and PNG. Gif and Jpeg are generally lossy
> compression, and don't fit the bill. PNG is a relatively new format, with
> good support from most OS/viewers/browsers. PNG is also licensed for use
> in a toolkit such as ITK. The real downside is that PNG only supports 8
> and 16 bit datatypes. I feel this is acceptable because medical image
> data generally comes in these two types, so that should not be a limiting
> factor. PNG support is built and working on my local checkout.
>
> Since I'm doing most of the testing infrastructure, I will offer to serve
> as the focal point for this issue, and welcome any suggestions.
>
> Thanks,
> -dan
>
> On Thu, 22 Mar 2001, Stephen R. Aylward wrote:
>
> >
> > How should we reference data files in the example directories if the
> > build is placing the executables in different directories? See below
> > for more details....
> >
> > Thanks,
> > Stephen
> >
> > Parag Chandra wrote:
> > >
> > > Ok, that solves half the problem. I'll put the test and the accompanying
> > > files into examples. But how would I reference that (the examples) directory from within
> > > the test program? Each platform will place the executable for this test into
> > > something like ${CMAKE_BINARY_DIR}/Examples, but ${CMAKE_BINARY_DIR} isn't
> > > even an environment variable for me to reference, it's a variable that's
> > > internal to CMake. If I reference the file without the pathname, like
> > > "TestInput.mhd", then the file has to be in the same directory as the
> > > executable, and that means I would have to place a copy of the test file
> > > into Insight-Cygwin, Insight-VC++, Insight-Linux, etc. Not only is that
> > > redundant, but there's no way I could foresee the additional platforms we
> > > might target, and so the test running on those platforms will generate an
> > > exception when it can't find the file. The only thing I can see is to place
> > > the test file into one directory, and then run the test for each platform
> > > from that directory by using the full pathname of the executable, e.g.
> > > /Insight-Cygwin/Examples/itkFileIOMetaImageTest.exe. Does anyone have any
> > > other ideas?
> > >
> > > Thanks,
> > > -Parag
> > >
> > > ---
> > > You are currently subscribed to caddlab as: aylward@unc.edu
> > > To unsubscribe send a blank email to leave-caddlab-706N@listserv.unc.edu
> >
> > --
> > ===============================================
> > Stephen R. Aylward
> > Assistant Professor of Radiology
> > Adjunct Assistant Professor of Computer Science
> > http://www.cs.unc.edu/~aylward
> > aylward@unc.edu
> > (919) 966-9695
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
> >
>
> --
> Daniel Blezek, Ph.D.
> blezek@crd.ge.com
> Visualization and Computer Vision Program
> Electronic Systems Lab
> GE Corporate Research & Development
--
===============================================
Stephen R. Aylward
Assistant Professor of Radiology
Adjunct Assistant Professor of Computer Science
http://www.cs.unc.edu/~aylward
aylward@unc.edu
(919) 966-9695
From chandra@cs.unc.edu Fri, 23 Mar 2001 14:03:36 -0500
Date: Fri, 23 Mar 2001 14:03:36 -0500
From: Parag Chandra chandra@cs.unc.edu
Subject: [Insight-developers] [Fwd: Changes to MetaImage...]
Daniel, I think this Toolkit::GetFileName() is exactly what we need, because
there has to be some way to get at the contents of variables like
${CMAKE_BINARY_DIR}. Otherwise the test process will have to cd to the
directory of the gold-standard images and then run each of the
platform-specific binaries via its fully-qualified pathname.
With regard to the file formats issue: since ImageIO is abstract, we have to
use a concrete reader for a specific file format in order to test. I chose
MetaImage because at one point I thought it was going to be one of the
"officially" supported formats. If this is no longer true, then we need to
decide on a format for testing purposes, and I think we need something more
comprehensive than PNG, or perhaps any other single file format for that
matter. 3D images are used frequently enough that I think 3D image I/O
should be part of the test process, but I realize 3D images would be
difficult to show in a browser for comparison. However, instead of a visual
comparison, couldn't we just use some sort of binary diff command?
-Parag
----- Original Message -----
From: "Daniel J. Blezek, Ph.D."
To: "Stephen R. Aylward"
Cc: "Insight-Developers (E-mail)"
Sent: Friday, March 23, 2001 8:24 AM
Subject: Re: [Insight-developers] [Fwd: Changes to MetaImage...]
> Stephen and Parag,
>
> This exact issue drove the discussion about file reading and writing at
> last week's telephone conference. I have several ideas in mind as to
> finding images and writing images for regression testing, but have not
> made them happen just yet. In the mean time, I would welcome any
> suggestions you may have as to how to address this issue. In our
> experience with VTK, which may be different for ITK, some platforms
> require a platform specific "gold image", and require special processing.
>
>
> Here are some of my ideas, strictly related to regression testing:
>
> - Have a Toolkit similar to Java, so to get the name of a file you would
> do something like this:
>
> std::string s = Toolkit::GetFileName ( "VHF/Head.png" );
>
> The Toolkit would properly expand the file name to be the path to the
> image you want, allowing images to be stored on fixed media(CDROM). On
> the outbound side, you would request an output image in a similar manner:
>
> std::string s = Toolkit::GetImageTestResultFileName (
> "itkFileIOMetaImageTest" );
>
> The Toolkit would expand this name to a directory where test images
> sit, and a latter process would pick up the images and compare them to the
> "gold standard". If the test were run in an example mode, it would simply
> write the output in the current directory, for the user to take a look at
> later.
>
>
> - This raises the issue of file formats yet again. Everyone seems to cook
> up their own file format to meet the needs of whatever problem they have
> at hand! (We are no exception...) In my view, the current problem at hand
> is regression testing. I've given this issue some thought. We need:
>
> - a file format that is easily displayed locally and through a browser
> - a format that is easily created by all Insight developers
> - a format that has 2D support at minimum
> - a format that supports as many common datatypes as possible
>
> The first point drives the selection down considerably. The only real
> canidates are Gif, Jpeg, and PNG. Gif and Jpeg are generally lossy
> compression, and don't fit the bill. PNG is a relatively new format, with
> good support from most OS/viewers/browsers. PNG is also licensed for use
> in a toolkit such as ITK. The real downside is that PNG only supports 8
> and 16 bit datatypes. I feel this is acceptable because medical image
> data generally comes in these two types, so that should not be a limiting
> factor. PNG support is built and working on my local checkout.
>
> Since I'm doing most of the testing infrastructure, I will offer to serve
> as the focal point for this issue, and welcome any suggestions.
>
> Thanks,
> -dan
>
>
> On Thu, 22 Mar 2001, Stephen R. Aylward wrote:
>
> >
> > How should we reference data files in the example directories if the
> > build is placing the executables in different directories? See below
> > for more details....
> >
> > Thanks,
> > Stephen
> >
> > Parag Chandra wrote:
> > >
> > > Ok, that solves half the problem. I'll put the test and the
accompanying
> > > files into examples. But how would I reference that (the examples)
directory from within
> > > the test program? Each platform will place the executable for this
test into
> > > something like ${CMAKE_BINARY_DIR}/Examples, but ${CMAKE_BINARY_DIR}
isn't
> > > even an environment variable for me to reference, it's a variable
that's
> > > internal to CMake. If I reference the file without the pathname, like
> > > "TestInput.mhd", then the file has to be in the same directory as the
> > > executable, and that means I would have to place a copy of the test
file
> > > into Insight-Cygwin, Insight-VC++, Insight-Linux, etc. Not only is
that
> > > redundant, but there's no way I could foresee the additional platforms
we
> > > might target, and so the test running on those platforms will generate
an
> > > exception when it can't find the file. The only thing I can see is to
place
> > > the test file into one directory, and then run the test for each
platform
> > > from that directory by using the full pathname of the executable, e.g.
> > > /Insight-Cygwin/Examples/itkFileIOMetaImageTest.exe. Does anyone have
any
> > > other ideas?
> > >
> > > Thanks,
> > > -Parag
> > >
> > > ---
> > > You are currently subscribed to caddlab as: aylward@unc.edu
> > > To unsubscribe send a blank email to
leave-caddlab-706N@listserv.unc.edu
> >
> > --
> > ===============================================
> > Stephen R. Aylward
> > Assistant Professor of Radiology
> > Adjunct Assistant Professor of Computer Science
> > http://www.cs.unc.edu/~aylward
> > aylward@unc.edu
> > (919) 966-9695
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
> >
>
> --
> Daniel Blezek, Ph.D.
> blezek@crd.ge.com
> Visualization and Computer Vision Program
> Electronic Systems Lab
> GE Corporate Research & Development
>
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
From brad.king@kitware.com Fri, 23 Mar 2001 16:37:07 -0500 (EST)
Date: Fri, 23 Mar 2001 16:37:07 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] Simplifying traits
Hello, all:
Here are some ideas to consider before deciding to throw out the use of
ScalarTraits an VectorTraits. Our current approach is not the simplest
approach to using traits.
We can make traits easier to use. Consider the following filter:
template
class MyFilter
{
public:
typedef TInputImage InputImageType;
typedef typename InputImageType::PixelType InputPixelType;
typedef ScalarTraits PixelScalarTraits;
typedef typename PixelScalarTraits::ScalarValueType
PixelScalarValueType;
inline static PixelScalarValueType& GetPixelScalar(PixelType& p)
{ return PixelScalarTraits::GetScalar(p); }
inline static void SetPixelScalar(PixelType& p,
const PixelScalarValueType& v)
{ PixelScalarTraits::SetScalar(p, v); }
// ....
};
Note that the function definitions can be put in ImageToImageFilter and be
inherited into other filters. Then, with minimal setup work, a filter
implementation can access a Pixel's scalar value like this:
PixelScalarValueType v = GetPixelScalar(myPixel);
// do something to v.
SetPixelScalar(myPixel, v);
This is in place of the current code:
typename ScalarTraits::ScalarValueType v =
ScalarTraits::GetScalar(myPixel);
// do something to v.
ScalarTraits::SetScalar(myPixel, v);
Also, I have noticed that there are a bunch of extra ScalarTrait
specializations for more than the basic types. An example is
"ScalarTraits< RGBPixel >". These should not be done, because if
someone wants to use an RGBPixel with a template argument other than those
defined in itkPixelTraits.h, they will have to figure out how to write
their own trait specialization for it. Currently, this can be solved (I
think) by putting the proper typedefs (ValueType, etc) into RGBPixel and
letting the primary template of ScalarTraits handle it from there.
The "real" way to do such a specialization of the ScalarTraits for
RGBPixel would be to do a partial specialization:
template
class ScalarTraits< RGBPixel > { /* ... */ };
Unfortunately, not all of our compilers support partial specialization, so
we can't do this.
The other option is to not require scalar traits to be a speicalization of
ScalarTraits at all. If instead we make such traits a template argument,
then any class can be defined and passed to the argument:
template >
class Image
{
public:
typedef TPixel PixelType;
enum { Dimension = VDimension };
typedef TPixelTraits PixelTraits;
// ....
};
Then, filters could pull the Pixel traits out of their image template
parameters:
template
class MyFilter
{
public:
typedef TInputImage InputImageType;
typedef typename InputImageType::PixelTraits InputPixelTraits;
// ... use traits as in previous MyFilter example.
};
Making traits a template argument is currently the approach that the Mesh
class uses. It even encapsulates the "CellTraits" as a member of the
"MeshTraits" template argument. We could make the same relationship for
the Image. That is:
template
class DefaultImageTraits
{
public:
typedef TPixel PixelType;
enum { Dimension = VDimension };
typedef DefaultPixelTraits PixelTraits;
// ... other useful traits?
};
template >
class Image
{
public:
typedef TImageTraits ImageTraits;
typedef typename ImageTraits::PixelType PixelType;
typedef typename ImageTraits::PixelTraits PixelTraits;
// ...
};
Assume that we have encapsulated both the current scalar and vector traits
into "DefaultPixelTraits".
Thoughts, anyone?
-Brad
From brad.king@kitware.com Fri, 23 Mar 2001 17:42:03 -0500 (EST)
Date: Fri, 23 Mar 2001 17:42:03 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] How to use "typename"
Hello, all:
90% of the compiler errors this week have been due to improper use of the
"typename" keyword. I would like to try to reduce this trend. This
message should help you understand when it should and should not be used.
There are actually two meanings of the "typename" keyword.
1.) Easy version:
When used to declare a template parameter, it is equivalent to
"class". This example demonstrates this case. Both of these are
identical:
template struct Foo {};
// is identical to
template struct Foo {};
2.) Hard version:
The keyword is used to tell the compiler that the next qualified
identifier will refer to a type. Rather than explaining the details of
why this is necessary, I will just give examples of most of the cases.
template
class MyClass
{
// "typename" not needed because there is no "::" access.
typedef T MyTemplateArgument;
// "typename" needed because "T" is a template parameter of MyClass.
typedef typename T::Foo Foo;
// "typename" not needed because Vector does not depend
// on a template parameter of MyClass.
typedef Vector::Iterator VectorIterator_int;
// "typename" needed because Vector DEPENDS on a template
// parameter of MyClass.
typedef typename Vector::Iterator VectorIterator_T;
// "typename" not needed because "Vector" is just a type.
typedef Vector FloatVector;
// "typename" not needed because "FloatVector" does not depend
// on a template paramaeter of MyClass.
typedef FloatVector::Iterator FloatVectorIterator;
// "typename" not needed because there is no "::" access.
typedef Vector MyVector;
// "typename" needed because "MyVector" depends on a template parameter
// of MyClass.
typedef typename MyVector::Iterator MyVectorIterator;
void MyMethod(); // used later in example
};
All of the above rules apply whenever you are referring to the types, not
just in a typedef. They are even needed when referring to a type as the
template argument for another class. However, once you have the typedef'd
name, you can just use that since the compiler knows it is a type.
For example:
template
void MyClass::MyMethod()
{
typename Vector::Iterator myIter;
// is equivalent to
typename MyVector::Iterator myIter;
// is equivalent to
MyVectorIterator myIter;
}
Whenever you are unsure about whether to use a typename, this email should
have an example of your situation. If not, feel free to email the list
with the situation, and I'm sure you'll get an answer quickly.
-Brad
From cates@cayenne.cs.utah.edu Fri, 23 Mar 2001 16:51:55 -0700 (MST)
Date: Fri, 23 Mar 2001 16:51:55 -0700 (MST)
From: Joshua Cates cates@cayenne.cs.utah.edu
Subject: [Insight-developers] pipeline update mechanism
Hello,
Has the Update() mechanism in the pipeline been modified? I am no longer
able to get filters to execute via Update() (as of today).
Josh.
______________________________
Josh Cates
School of Computer Science
University of Utah
Email: cates@cs.utah.edu
Phone: (801) 587-7697
URL: www.cs.utk.edu/~cates
From ibanez@cs.unc.edu Fri, 23 Mar 2001 18:25:28 -0500
Date: Fri, 23 Mar 2001 18:25:28 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] itkMatrix
Hi,
itkMatrix has been reworked.
it doesn't derive from vnl_matrix_fixed anymore,
now it contains a vnl_matrix_fixed.
The product operations are defined for :
- itkPoint
- itkVector
- itkCovariantVector
- itkMatrix
- vnl_matrix
- vnl_vector
Access to the internal vnl_matrix can be get
through GetVnlMatrix().
it can be initalized with
SetIdentity();
and accept assigments from
itkMatrix
vnl_matrix
Luis
From lorensen@crd.ge.com Mon, 26 Mar 2001 07:52:46 -0500
Date: Mon, 26 Mar 2001 07:52:46 -0500
From: Lorensen, William E (CRD) lorensen@crd.ge.com
Subject: [Insight-developers] IMPORTANT: pre-commit checking for cvs
Folks,
I've added pre-commit check scripts to the cvs repository. Currently there are two pre-commit
checks.
Both operate on txx, cxx, h, htm and html files.
pre-commit check #1 scans the file for lines with a trailing ctrl-M. These are probably being
inserted by Micorsoft
editors. They are easliy removed with emacs. I'm not sure how to remove them otherwise.
pre-commit check #2 checks the last line in a file for a newline. This is easliy fixed with any
editor. Go to the end of the file and hit enter.
If either pre-commit check fails, the entire cvs commit wiill be stopped.
I hope this will help folks.
I'm not sure about the robustness of these scripts. Please send me e-mail if you try to commit a file
and the checkers incorrectly block the commit.
Bill
From ibanez@cs.unc.edu Mon, 26 Mar 2001 10:15:56 -0500
Date: Mon, 26 Mar 2001 10:15:56 -0500
From: Luis Ibanez ibanez@cs.unc.edu
Subject: [Insight-developers] Dashboard / Registration
Hi,
We are changing the part of the API in the registration
framework. Previously a VectorContainer was used to
pass the parameters of the spatial transformation to the
optimizers. This was impractical because VectorContainer
elements has to be accessed through iterators.
So, now the framework is templated the parameters type,
and it could be any type with the operator[]() defined.
The bad news is that the changes are not completed yet
so there are a lot of new error messages on the dashboard,
coming from these modifications. We are working now
on fixing them....
Luis
From blezek@crd.ge.com Mon, 26 Mar 2001 10:38:07 -0500 (EST)
Date: Mon, 26 Mar 2001 10:38:07 -0500 (EST)
From: Daniel J. Blezek, Ph.D. blezek@crd.ge.com
Subject: [Insight-developers] Image file formats for testing
Parag and Stephen,
Before my life at GE, I worked with Dr. Rick Robb at Mayo on Analyze.
Again, the Analyze file format is simply a "roll-your-own" format, that
served the need of the group at the time. The current incarnation of
Analyze, AVW has a better model of a file format. The file is simply text
indicating spacing, scanner, datatype, etc... with filenames to read as
each slice. Thus, the individual slices are 2D but wrapped into 3D.
I'd like to make it clear that I'm not proposing a fileformat for all of
Insight to use forever, just a format to settle on for the purpose of
testing our code. I still feel that PNG makes sense. The specification
allows you to set the physical scale units(In Meters), and height and
width, and we can encode depth into the text comments that PNG allows.
I'm really more interested in PNG as an _output_ format, not necessarly as
an input format, e.g. a test reads whatever format it likes (assuming it's
a cross-platform library), does it's processing, and writes out a PNG
file. A later process picks up the PNG files, compares them to the "gold
standard" for that algorithm, producing difference images if the test
generated a different result. This would allow a developer to produce
input data in whatever (supported) format is most convient, and allow
everyone to view the testing output, and for the dashboard to easily show
the images.
On a related note, I've been considering a framework to allow Insight to
be more easily streamlined with other packages, sort of an import/export
facility, with a simplified pipeline interface. That would make using
Insight with VTK or other packages straightforward. In this design, it
makes sense to move the ImageIO class under the import/export classes, for
after all, file i/o is interaction with an external system. This should
be relatively straightforward, and not very invasive, but will involve
re-arranging the code a bit. I'm hoping to get some time to work on this
later this week, and I'll run my design by you both, before making changes
to CVS.
Thanks,
-dan
> The only difficulty is that PNG doesn't provide element spacing
> information. How about a medical format - e.g., analyze?
> Stephen
On Fri, 23 Mar 2001, Parag Chandra wrote:
> Daniel, I think this Toolkit::GetFileName() is exactly what we need, because
> there has to be some way to get at the contents of variables like
> ${CMAKE_BINARY_DIR}. Otherwise the test process will have to cd to the
> directory of the gold-standard images and then run each of the
> platform-specific binaries via its fully-qualified pathname.
>
> With regard to the file formats issue: since ImageIO is abstract, we have to
> use a concrete reader for a specific file format in order to test. I chose
> MetaImage because at one point I thought it was going to be one of the
> "officially" supported formats. If this is no longer true, then we need to
> decide on a format for testing purposes, and I think we need something more
> comprehensive than PNG, or perhaps any other single file format for that
> matter. 3D images are used frequently enough that I think 3D image I/O
> should be part of the test process, but I realize 3D images would be
> difficult to show in a browser for comparison. However, instead of a visual
> comparison, couldn't we just use some sort of binary diff command?
>
> -Parag
>
> ----- Original Message -----
> From: "Daniel J. Blezek, Ph.D."
> To: "Stephen R. Aylward"
> Cc: "Insight-Developers (E-mail)"
> Sent: Friday, March 23, 2001 8:24 AM
> Subject: Re: [Insight-developers] [Fwd: Changes to MetaImage...]
>
>
> > Stephen and Parag,
> >
> > This exact issue drove the discussion about file reading and writing at
> > last week's telephone conference. I have several ideas in mind as to
> > finding images and writing images for regression testing, but have not
> > made them happen just yet. In the mean time, I would welcome any
> > suggestions you may have as to how to address this issue. In our
> > experience with VTK, which may be different for ITK, some platforms
> > require a platform specific "gold image", and require special processing.
> >
> >
> > Here are some of my ideas, strictly related to regression testing:
> >
> > - Have a Toolkit similar to Java, so to get the name of a file you would
> > do something like this:
> >
> > std::string s = Toolkit::GetFileName ( "VHF/Head.png" );
> >
> > The Toolkit would properly expand the file name to be the path to the
> > image you want, allowing images to be stored on fixed media(CDROM). On
> > the outbound side, you would request an output image in a similar manner:
> >
> > std::string s = Toolkit::GetImageTestResultFileName (
> > "itkFileIOMetaImageTest" );
> >
> > The Toolkit would expand this name to a directory where test images
> > sit, and a latter process would pick up the images and compare them to the
> > "gold standard". If the test were run in an example mode, it would simply
> > write the output in the current directory, for the user to take a look at
> > later.
> >
> >
> > - This raises the issue of file formats yet again. Everyone seems to cook
> > up their own file format to meet the needs of whatever problem they have
> > at hand! (We are no exception...) In my view, the current problem at hand
> > is regression testing. I've given this issue some thought. We need:
> >
> > - a file format that is easily displayed locally and through a browser
> > - a format that is easily created by all Insight developers
> > - a format that has 2D support at minimum
> > - a format that supports as many common datatypes as possible
> >
> > The first point drives the selection down considerably. The only real
> > canidates are Gif, Jpeg, and PNG. Gif and Jpeg are generally lossy
> > compression, and don't fit the bill. PNG is a relatively new format, with
> > good support from most OS/viewers/browsers. PNG is also licensed for use
> > in a toolkit such as ITK. The real downside is that PNG only supports 8
> > and 16 bit datatypes. I feel this is acceptable because medical image
> > data generally comes in these two types, so that should not be a limiting
> > factor. PNG support is built and working on my local checkout.
> >
> > Since I'm doing most of the testing infrastructure, I will offer to serve
> > as the focal point for this issue, and welcome any suggestions.
> >
> > Thanks,
> > -dan
> >
> >
> > On Thu, 22 Mar 2001, Stephen R. Aylward wrote:
> >
> > >
> > > How should we reference data files in the example directories if the
> > > build is placing the executables in different directories? See below
> > > for more details....
> > >
> > > Thanks,
> > > Stephen
> > >
> > > Parag Chandra wrote:
> > > >
> > > > Ok, that solves half the problem. I'll put the test and the
> accompanying
> > > > files into examples. But how would I reference that (the examples)
> directory from within
> > > > the test program? Each platform will place the executable for this
> test into
> > > > something like ${CMAKE_BINARY_DIR}/Examples, but ${CMAKE_BINARY_DIR}
> isn't
> > > > even an environment variable for me to reference, it's a variable
> that's
> > > > internal to CMake. If I reference the file without the pathname, like
> > > > "TestInput.mhd", then the file has to be in the same directory as the
> > > > executable, and that means I would have to place a copy of the test
> file
> > > > into Insight-Cygwin, Insight-VC++, Insight-Linux, etc. Not only is
> that
> > > > redundant, but there's no way I could foresee the additional platforms
> we
> > > > might target, and so the test running on those platforms will generate
> an
> > > > exception when it can't find the file. The only thing I can see is to
> place
> > > > the test file into one directory, and then run the test for each
> platform
> > > > from that directory by using the full pathname of the executable, e.g.
> > > > /Insight-Cygwin/Examples/itkFileIOMetaImageTest.exe. Does anyone have
> any
> > > > other ideas?
> > > >
> > > > Thanks,
> > > > -Parag
> > > >
> > > > ---
> > > > You are currently subscribed to caddlab as: aylward@unc.edu
> > > > To unsubscribe send a blank email to
> leave-caddlab-706N@listserv.unc.edu
> > >
> > > --
> > > ===============================================
> > > Stephen R. Aylward
> > > Assistant Professor of Radiology
> > > Adjunct Assistant Professor of Computer Science
> > > http://www.cs.unc.edu/~aylward
> > > aylward@unc.edu
> > > (919) 966-9695
> > >
> > > _______________________________________________
> > > Insight-developers mailing list
> > > Insight-developers@public.kitware.com
> > > http://public.kitware.com/mailman/listinfo/insight-developers
> > >
> >
> > --
> > Daniel Blezek, Ph.D.
> > blezek@crd.ge.com
> > Visualization and Computer Vision Program
> > Electronic Systems Lab
> > GE Corporate Research & Development
> >
> >
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
> >
>
--
Daniel Blezek, Ph.D.
blezek@crd.ge.com
Visualization and Computer Vision Program
Electronic Systems Lab
GE Corporate Research & Development
From hughett@mercur.uphs.upenn.edu Mon, 26 Mar 2001 12:28:09 -0500
Date: Mon, 26 Mar 2001 12:28:09 -0500
From: Paul Hughett hughett@mercur.uphs.upenn.edu
Subject: [Insight-developers] Build broken on Linux
I just did an update, make clean, configure, and make on Linux (RH
6.2) using an in-source build. I get the following fatal error (after
bypassing a couple of optimization tests that won't compile):
make[4]: Entering directory `/home/hughett/work/Insight/Testing/Code/Common'
/home/hughett/work/Insight/CMake/Source/CMakeBuildTargets `cd .; pwd`/CMakeLists.txt -S`cd .; pwd` -O`pwd` -H/home/hughett/work/Insight -B/home/hughett/work/Insight
CMake Error: error cannot find dependencies for itkTransform.h
CMake Error: error cannot find dependencies for itkTransform.h
make[4]: *** [CMakeTargets.make] Error 255
Paul Hughett
From lng@insightful.com Mon, 26 Mar 2001 12:12:08 -0800
Date: Mon, 26 Mar 2001 12:12:08 -0800
From: Lydia Ng lng@insightful.com
Subject: [Insight-developers] FW: VC++ memccpy warnings
Hi,
The following VC++ warnings (see below) are due to inclusion of
#include
instead of
#include
Are there any reasons why we can't uniformly use the second method?
This change should get rid of about 20+ warnings for VC++.
Files using the first method are:
\CODE\Common\itkImageConstIteratorWithIndex.h(20):#include
\CODE\Common\itkImageIteratorWithIndex.h(20):#include
\CODE\Common\itkScalarVector.h(21):#include
Cheers,
Lydia
------------------------------------------
Compiling...
itkMutualInformationMetricTest.cxx
C:\Program Files\Microsoft Visual Studio\VC98\INCLUDE\memory.h(68) : warning
C4273: '_memccpy' : inconsistent dll linkage. dllexport assumed.
C:\Program Files\Microsoft Visual Studio\VC98\INCLUDE\memory.h(69) : warning
C4273: 'memchr' : inconsistent dll linkage. dllexport assumed.
C:\Program Files\Microsoft Visual Studio\VC98\INCLUDE\memory.h(70) : warning
C4273: '_memicmp' : inconsistent dll linkage. dllexport assumed.
C:\Program Files\Microsoft Visual Studio\VC98\INCLUDE\memory.h(85) : warning
C4273: 'memccpy' : inconsistent dll linkage. dllexport assumed.
C:\Program Files\Microsoft Visual Studio\VC98\INCLUDE\memory.h(86) : warning
C4273: 'memicmp' : inconsistent dll linkage. dllexport assumed.
Linking...
Creating library Release/itkMutualInformationMetricTest.lib and object
Release/itkMutualInformationMetricTest.exp
From aylward@unc.edu Mon, 26 Mar 2001 16:02:39 -0500
Date: Mon, 26 Mar 2001 16:02:39 -0500
From: Stephen R. Aylward aylward@unc.edu
Subject: [Insight-developers] IMPORTANT: pre-commit checking for cvs
Hi,
If you want - via anonymous ftp, there is a 91K zip file at
ftp.rad.unc.edu/pub/users/aylward
called
tofrodos.zip
The zip archive contains source and executable for linux, unix, win, and
dos for converting files to/from dos/unix.
I tested - it will handle the case of only a few lines having ctrl-m's
on the end.
A good set of options is
todos -fodp
-f force overwrite
-o overwrite original
-d convert dos to unix
-p preserve file owner and time (important for CVS)
Stephen
"Lorensen, William E (CRD)" wrote:
>
> Folks,
> I've added pre-commit check scripts to the cvs repository. Currently there are two pre-commit
> checks.
> Both operate on txx, cxx, h, htm and html files.
>
> pre-commit check #1 scans the file for lines with a trailing ctrl-M. These are probably being
> inserted by Micorsoft
> editors. They are easliy removed with emacs. I'm not sure how to remove them otherwise.
>
> pre-commit check #2 checks the last line in a file for a newline. This is easliy fixed with any
> editor. Go to the end of the file and hit enter.
>
> If either pre-commit check fails, the entire cvs commit wiill be stopped.
>
> I hope this will help folks.
>
> I'm not sure about the robustness of these scripts. Please send me e-mail if you try to commit a file
> and the checkers incorrectly block the commit.
>
> Bill
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
--
===============================================
Stephen R. Aylward
Assistant Professor of Radiology
Adjunct Assistant Professor of Computer Science
http://www.cs.unc.edu/~aylward
aylward@unc.edu
(919) 966-9695
From blezek@crd.ge.com Mon, 26 Mar 2001 16:18:33 -0500 (EST)
Date: Mon, 26 Mar 2001 16:18:33 -0500 (EST)
From: Daniel J. Blezek, Ph.D. blezek@crd.ge.com
Subject: [Insight-developers] Build broken on Linux
Paul,
I find that if I re-run make, this clears up.
-dan
On Mon, 26 Mar 2001, Paul Hughett wrote:
>
> I just did an update, make clean, configure, and make on Linux (RH
> 6.2) using an in-source build. I get the following fatal error (after
> bypassing a couple of optimization tests that won't compile):
>
> make[4]: Entering directory `/home/hughett/work/Insight/Testing/Code/Common'
> /home/hughett/work/Insight/CMake/Source/CMakeBuildTargets `cd .; pwd`/CMakeLists.txt -S`cd .; pwd` -O`pwd` -H/home/hughett/work/Insight -B/home/hughett/work/Insight
> CMake Error: error cannot find dependencies for itkTransform.h
> CMake Error: error cannot find dependencies for itkTransform.h
> make[4]: *** [CMakeTargets.make] Error 255
>
>
> Paul Hughett
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>
--
--
Daniel Blezek, Ph.D.
blezek@crd.ge.com
Visualization and Computer Vision Program
Electronic Systems Lab
GE Corporate Research & Development
From zhuge@mipg.upenn.edu Mon, 26 Mar 2001 16:42:28 -0600
Date: Mon, 26 Mar 2001 16:42:28 -0600
From: Ying Zhuge zhuge@mipg.upenn.edu
Subject: [Insight-developers] Check in
Hi,
I want to check in our VectorFuzzyConnectednessImageFilter, but I fail.
The problem is no repository.
I use wincvs under windows 98, the command is:
cvs add myfile
How can I do? Thanks
Ying
From chenting@graphics.cis.upenn.edu Mon, 26 Mar 2001 15:58:35 -0600
Date: Mon, 26 Mar 2001 15:58:35 -0600
From: Ting Chen chenting@graphics.cis.upenn.edu
Subject: [Insight-developers] the warning message
hi! Guys,
did any of you into the following warning message?
vc98\include\xmemory(64) : warning C4786:
'std::allocator,itk:
:VectorContainer >,std::set,std::allocator > > > > >' :
identifier was truncated to '255' characters in the debug information
d:\program files\microsoft visual studio\vc98\include\xmemory(64) :
while compiling class-template member function 'void __thiscall
std::allocator,itk::VectorContainer >,std::set,std::allocator > > > > >::deallocate(void *,unsigned
int)'
how to avoid it?
Thanks!
ting
-----Original Message-----
From: Ying Zhuge
To: insight-developers@public.kitware.com
Date: Monday, March 26, 2001 3:42 PM
Subject: [Insight-developers] Check in
>Hi,
>I want to check in our VectorFuzzyConnectednessImageFilter, but I fail.
>The problem is no repository.
>I use wincvs under windows 98, the command is:
> cvs add myfile
>
>How can I do? Thanks
>
>Ying
>
>
>_______________________________________________
>Insight-developers mailing list
>Insight-developers@public.kitware.com
>http://public.kitware.com/mailman/listinfo/insight-developers
From lorensen@crd.ge.com Mon, 26 Mar 2001 16:50:43 -0500
Date: Mon, 26 Mar 2001 16:50:43 -0500
From: Lorensen, William E (CRD) lorensen@crd.ge.com
Subject: [Insight-developers] Build broken on Linux
Remove CMakeTargets.make and Cxx/*.cxx
Then remake
-----Original Message-----
From: Paul Hughett [mailto:hughett@mercur.uphs.upenn.edu]
Sent: Monday, March 26, 2001 12:28 PM
To: insight-developers@public.kitware.com
Subject: [Insight-developers] Build broken on Linux
I just did an update, make clean, configure, and make on Linux (RH
6.2) using an in-source build. I get the following fatal error (after
bypassing a couple of optimization tests that won't compile):
make[4]: Entering directory `/home/hughett/work/Insight/Testing/Code/Common'
/home/hughett/work/Insight/CMake/Source/CMakeBuildTargets `cd .; pwd`/CMakeLists.txt -S`cd .; pwd`
-O`pwd` -H/home/hughett/work/Insight -B/home/hughett/work/Insight
CMake Error: error cannot find dependencies for itkTransform.h
CMake Error: error cannot find dependencies for itkTransform.h
make[4]: *** [CMakeTargets.make] Error 255
Paul Hughett
_______________________________________________
Insight-developers mailing list
Insight-developers@public.kitware.com
http://public.kitware.com/mailman/listinfo/insight-developers
From brad.king@kitware.com Mon, 26 Mar 2001 17:10:36 -0500 (EST)
Date: Mon, 26 Mar 2001 17:10:36 -0500 (EST)
From: Brad King brad.king@kitware.com
Subject: [Insight-developers] the warning message
> did any of you into the following warning message?
>
> vc98\include\xmemory(64) : warning C4786:
> 'std::allocator