[IGSTK-Developers] LOG4CXX test on IGSTK state machine

[Luis Ibanez]

Kevin,

We didn't talked about log4cxx at the Tcon since we were focused
on the issues related to the integration of David's tracker code.

As Julien pointed out, the current insertion of log4cxx into ITK
have faced a number of technical difficulties that can be summarized
as


1) Multi platform limitations.
    It seems that it will be difficult to adapt
    to all the platforms that are supported in ITK.

2) Maintenance for a discontinued version.
    The version we put in ITK has been discontinued
    by Apache. In their new version they depend on
    another library for system-level services.

3) Creation of singletons per Class.
    Bill Hoffman (at Kitware) pointed out that
    having singletons per class have been attempted
    in VTK in the past, and have proved to be unfit
    for many of the use cases of a library.
    In particular when using DLL and at the moment of
    quiting an application.  The circumstances that
    Bill described will compromise the reliability of
    ITK and therefore IGSTK for being used in a critical
    environment such as the Operating Room


We discussed these issues with Jim Miller and Daniel Blezeck
at GE and agreed in following the approach that Sohan has
developed so far. Meaning, studying the functionalities of
Log4cxx and adapting them to a set of classes the we
design ourselves and can model to our needs.

Our current action item is to gather the requirements of the
logging class in the context of ITK and put those requirements
in the ITK Wiki.

If Hee-su Kim is available for further developing this code,
that will be great. We still want to push for integrating the
logging capabilities into ITK, but we will not pursue the use
of log4cxx as a direct third party library.



    Luis




--------------------
Kevin Cleary wrote:
> Julien and others:
> 
> This is actually a very valuable update as this was not mentioned on the
> Thursday tcon (and since it was only about tracking you did not miss
> anything)
> 
> It does sound like we may need to be ready to do our own logging code and I
> will probably have Hee-su Kim from our group be responsible for this - for
> right now I will continue to have him look at log4cxx as an example and I
> did put logging on the agenda for the advisory board meeting (we might not
> want to wait for that)
> 
> Kevin
> 
> P.S. to Julien: I know I still owe you some info on hotels which the
> secretary has not given me but I will be back at Georgetown on Monday and
> will track it down then
> 
> -----Original Message-----
> From: igstk-developers-bounces at public.kitware.com
> [mailto:igstk-developers-bounces at public.kitware.com] On Behalf Of Julien
> Jomier
> Sent: Friday, March 18, 2005 12:17 PM
> To: Hee-su Kim
> Cc: igstk-developers at public.kitware.com
> Subject: Re: [IGSTK-Developers] LOG4CXX test on IGSTK state machine
> 
> Hi,
> 
> Sorry I was not on the T-Con yesterday. I don't know if Luis was...
> Just a couple of things on log4cxx since we tried to add it into ITK 
> last week with some difficulties.
> 
> 1) Only the "old" release of log4cxx can be used since the new one 
> requires the Apache package. That means that we won't be able to update 
> log4cxx in the future.
> 
> 2) I've worked on fixing the "old" release to compile with cygwin (gcc), 
> Borland compiler and Intel compilers. However log4cxx currently does not 
> support gcc 2.95 and it is not a trivial thing to make it work on this 
> compiler.
> 
> That said, log4cxx will probably be removed from ITK, unless the gcc 
> 2.95 and other issues are solved. At the same time, Luis is collecting 
> some ideas/feature requests to design our own logging facility. (Luis 
> might be able to give more info on that).
> Therefore, I'm not sure if we should still rely on log4cxx for IGSTK. 
> IMHO, we only need basic logging for IGSTK (smtp, socket, etc... are not 
> useful).
> 
> Sorry if Luis as already pointed this out yesterday,
> 
> Julien
> 
> Hee-su Kim wrote:
> 
>>Hi,
>> 
>>I compiled LOG4CXX with ITK. (I added LOG4CXX source codes into ITK 
>>project and then built it.)
>> 
>>LOG4CXX codes were used for logging state machine error messages and for 
>>each transition made during running.
>>(instead of std::cerr)
>> 
>>Attached are screenshots from IGSTK state machine errors test program.
>> 
>>Some issues here:
>>    1. Should LOG4CXX Logger be derived from igstkLogger class?
>>    2. igstkMultipleOutputs class seems not necessary if LOG4CXX is used.
>>    3. After ITK has integrated LOG4CXX, I think it should be used. 
>>Logging codes might have be changed then.
>>    4. Platform compatibility issue. I only built on MSVC6.
>> 
>>Hee-Su
>>
>>------------------------------------------------------------------------
>>
>>
>>------------------------------------------------------------------------
>>
>>
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>IGSTK-Developers mailing list
>>IGSTK-Developers at public.kitware.com
>>http://public.kitware.com/cgi-bin/mailman/listinfo/igstk-developers
> 
> 
> _______________________________________________
> IGSTK-Developers mailing list
> IGSTK-Developers at public.kitware.com
> http://public.kitware.com/cgi-bin/mailman/listinfo/igstk-developers
> 
> 
> 
> _______________________________________________
> IGSTK-Developers mailing list
> IGSTK-Developers at public.kitware.com
> http://public.kitware.com/cgi-bin/mailman/listinfo/igstk-developers
> 
>