[Insight-users] Re: itk java wrapper on cygwin problem (Progress)

Michael Bell michael.bell at acm.org
Wed Jul 26 09:59:55 EDT 2006


Gaetan,

I am using ext3 on linux, which has a maximum filesize of 255
characters. I see that Brad has addressed possible solutions to this.

I commented out the MetaDataDictionary, and the rest of the compile
finished without errors. Things seem to run well. I noticed that some
classes are named differently than the straight itk wrapping. For
example: itkImageMetricIF3 instead of itkImageMetricF3. Does the I
stand for Image?

Thanks for your help,
michael

On 7/25/06, Gaëtan Lehmann <gaetan.lehmann at jouy.inra.fr> wrote:
>
> Hi,
>
> I also have the same file with a very long name in my build tree, but I
> don't have any problem to access it. I thought this kind of problem was
> limited to windows (I say that but André didn't report any problem like
> that). What is you file system ?
> This kind of problem is already know and has been fixed for string lists
> and vectors in files Module/Base/*SwigExtras*.
> However, this workaround is inherited from the current ITK wrappers, and
> has not been modified in WrapITK - I'm not sure how it can be used to fix
> your problem.
>
> Brad, can you help on that point ?
>
> As a quick fix, I think you should disable the class which use this type:
> MetaDataDictionary.
> To do that, just comment the line
>
>    WRAP_NON_TEMPLATE_CLASS("itk::MetaDataDictionary")
>
> in the file Modules/Base/wrap_ITKCommonBase.cmake
> Taht being said, I think you will have the same problem with other types
> used in more useful classes :-/
> Let me now if it still fail (or if it work !)
>
> Gaetan
>
>
> Le Tue, 25 Jul 2006 15:55:29 +0200, Michael Bell <michael.bell at acm.org> a
> écrit:
>
> > Gaetan,
> >
> > I am trying wrapitk on linux while waiting to receive VS2003. I really
> > like the setup, especially being able to choose what is wrapped.
> >
> > I am getting the following error while compiling (ITK&CableSwig 2.8):
> > Generating wrap_ITKCommonBaseJava.cxx
> > Unable to open
> > /data/WrapITK/Build_Linux/Java/InsightToolkit/SWIGTYPE_p_std___Rb_tree_iteratorTstd__pairTconst_std__basic_stringTchar_std__char_traitsTchar_t_std__allocatorTchar_t_t_itk__SmartPointerTitk__MetaDataObjectBase_t_t_std__pairTconst_std__basic_stringTchar_std__char_traitsTchar_t_std__allocatorTchar_t_t_itk__SmartPointerTitk__MetaDataObjectBase_t_tR_std__pairTconst_std__basic_stringTchar_std__char_traitsTchar_t_std__allocatorTchar_t_t_itk__SmartPointerTitk__MetaDataObjectBase_t_tp_t.java
> >
> > The file name is too long. Any ideas on how to correct this?
> >
> > thanks,
> > michael
> >
> > On 7/19/06, Gaëtan Lehmann <gaetan.lehmann at jouy.inra.fr> wrote:
> >>
> >> Michael,
> >>
> >> André Bongers reported to be able to build WrapITK with the .NET 2003
> >> c++.
> >> WrapITK surely still have some bugs on windows plateform, but if you
> >> want
> >> to try it, I will surely be more reactive that official ITK developers.
> >> WrapITK can be downloaded at
> >>
> >>    http://voxel.jouy.inra.fr/darcs/contrib-itk/WrapITK/WrapITK.tar.gz
> >>
> >> or directly with darcs
> >>
> >>    darcs get --partial
> >> http://voxel.jouy.inra.fr/darcs/contrib-itk/WrapITK/
> >>
> >> and requires ITK 2.8.1
> >>
> >> Gaetan
> >>
> >>
> >> Le Wed, 19 Jul 2006 22:10:55 +0200, Michael Bell <michael.bell at acm.org>
> >> a
> >> écrit:
> >>
> >> > SwigInc.txt is being generated from SwigInc.txt.in, which has the
> >> > extra carriage return character. I suspect this is a problem because I
> >> > am using the cygwin recommended unix files, but this file is a dos
> >> > file. If I dos2unix SwigInc.txt.in and make, this error goes away, and
> >> > all of the SwigIn.txt files are correct. If I run dos2unix on the
> >> > entire Wrapping directory, many problems are fixed.
> >> >
> >> > Which shows the next issue:
> >> > javac does not understand cygwin paths. This issue can be fixed using
> >> > cygpath.
> >> >
> >> > Which shows the next problem:
> >> > javac VXLNumericsJava.java
> >> > cannot resolve symbol itkbase
> >> >
> >> > I haven't solved that problem, but I am certain it would reveal more.
> >> >
> >> > It's pretty clear that no one is maintaining the wrappers with cygwin.
> >> > Please make this clear somewhere in the documentation so the next
> >> > person does not waste weeks on this, as I have.
> >> >
> >> > So, we know that VS8 does not work. Cygwin does not work. What does
> >> > work on Windows? I can change compilers, but I can't spend time
> >> > hunting for a platform that does work. Please advise.
> >> >
> >> > thanks,
> >> > michael
> >> >
> >> >
> >> >
> >> >
> >> > On 7/17/06, Michael Bell <michael.bell at acm.org> wrote:
> >> >> ITK people,
> >> >>
> >> >> I have made some progress tracking down this error:
> >> >> wrap_vnl_matrix.xml from gccxml_cc1plus:
> >> >>  error: too many filenames given.  Type gccxml_cc1plus --help for
> >> usage
> >> >> make[2]: *** [Wrapping/CSwig/VXLNumerics/wrap_vnl_matrix.xml] Error 1
> >> >>
> >> >> This is the operation that fails (paths shortened for readability):
> >> >> gccxml -fxml-start=_cable_ -fxml=wrap_vnl_matrix.xml
> >> >> --gccxml-gcc-options SwigInc.txt -DCSWIG -DCABLE_CONFIGURATION
> >> >> wrap_vnl_matrix.cxx
> >> >>
> >> >> SwigInc.txt contains a list of include paths:
> >> >> -I/cygdrive/d/Applications/InsightToolkit-2.8.1/Build_Cygwin
> >> >> -I/cygdrive/d/Applications/InsightToolkit-2.8.1/Code/Algorithms
> >> >> ...
> >> >> -I/cygdrive/d/Applications/InsightToolkit-2.8.1/Code/BasicFilters
> >> >> ^M
> >> >>
> >> >> Then, it has the ^M character at the end. If I remove this character,
> >> >> the xml file appears to be created correctly. Does anyone know where
> >> >> this is being created, or why, or how I can prevent it? Is this an
> >> ITK
> >> >> problem, or a CableSwig problem?
> >> >>
> >> >> Please, I'm obviously willing to do the leg work, but it will take me
> >> >> 10 times as long to figure this out than if an ITK developer would
> >> >> step up and help. I've already shown that I was not the first person
> >> >> to deal with this problem, and there are likely others who have not
> >> >> said anything.
> >> >>
> >> >> thanks,
> >> >> michael
> >> >>
> >> >>
> >> >> --
> >> >> michael.bell at acm.org
> >> >>
> >> >
> >> >
> >>
> >>
> >>
> >> --
> >> Utilisant le client e-mail révolutionnaire d'Opera :
> >> http://www.opera.com/mail/
> >>
> >
> >
>
>
>
> --
> Utilisant le client e-mail révolutionnaire d'Opera :
> http://www.opera.com/mail/
>


-- 
michael.bell at acm.org


More information about the Insight-users mailing list