[Insight-developers] Re: STEP 1: itpack.cxx: moved

[Lorensen, William E (Research)]

Looks like all the FEM tests are crashing on Borland with a floating point
overflow.

-----Original Message-----
From: Luis Ibanez [mailto:luis . ibanez at kitware . com]
Sent: Friday, August 08, 2003 1:04 AM
To: Lorensen, William E (Research)
Cc: Brian B Avants; tessa; Aljaz Noe; Insight Developers List;
gee at rad . upenn . edu; jtduda at grasp . cis . upenn . edu
Subject: Re: [Insight-developers] Re: STEP 1: itpack.cxx: moved



Bill,

It seems to be ok now.

itpack.h is now included only in the .cxx
FEM wrappers for the sparse matrix and the
linear solver.

The f2c.h types exposed in the API were
replaced by their equivalents (int, double).

A couple of fixes were made also in the
itkFEMRegistrationFilter.h, txx.

It is building fine now on gcc 3.02 and
VC++ Net.

The FEMAtlasSegmentation application
in InsightApplications/IBSRValidation
is building ok too.



   Luis



---------------------
Luis Ibanez wrote:
> more hints...
> 
> 
> itkFEMItpackSparseMatrix.h
> itkFEMLinearSystemWrapperItpack.h
> itkFEMSolverCrankNicolson.h
> 
> include itpack.h,
> 
> itpack.h includes f2c.h
> 
> and f2c.h has a lot of nasty exposed code like
> 
>>
>> #define abs(x) ((x) >= 0 ? (x) : -(x))
>> #define dabs(x) (doublereal)abs(x)
>> #define min(a,b) ((a) <= (b) ? (a) : (b))
>> #define max(a,b) ((a) >= (b) ? (a) : (b))
>> #define dmin(a,b) (doublereal)min(a,b)
>> #define dmax(a,b) (doublereal)max(a,b)
>>
> 
> ....
> 
> the f2c.h should only be exposed to the .cxx
> files that are wrapping itpack.
> 
> ... working on it...
> 
> Luis
> 
> 
> 
> ----------------------------------------------
> Lorensen, William E (Research) wrote:
> 
>> Luis,
>> Some strange side effect from someones changes. Any ideas?
>>
>> Bill
>>