Hi Ben
scons can do that very easily (platform specific checks). That's why I
want to move from distutils to scons build system as soon as possible.
Thomas
ben lipkowitz a écrit :
> since the directory may or may not always exist on any given linux
> distribution, perhaps a better solution is to check to see if it exists?
> like
>
> if os.access('/usr/include/', os.F_OK):
> OCC_INC = '/usr/include/'
>
> etc
>
> there are not any directories named "inc" on my debian installation
>
> On Sun, 15 Mar 2009, Arthur Magill wrote:
>
>> Date: Sun, 15 Mar 2009 16:37:56 +0100
>> From: Arthur Magill <arthur.mag...@epfl.ch>
>> To: Thomas Paviot <thomas.pav...@free.fr>
>> Cc: "pythonocc-users@gna.org" <pythonocc-users@gna.org>
>> Subject: Re: [Pythonocc-users] New modules for Linux/Darwin
>>
>> Hi Thomas,
>>
>>
>>> To be as portable as possible, it's better to use the os.path.join
>>> method:
>>> OCC_INC = os.path.join(OCC_ROOT,'inc')
>>> OCC_LIB = os.path.join(OCC_ROOT,'lib')
>>
>> Yep, that's certainly better.
>>
>>>
>>>> in environment.py? I think that should work on all Linux
>>>> installations.
>>>
>>> As far as I know, according to the OpenCascade package you choose to
>>> install (whether it is under Ubuntu or OpenSuse), /inc and /lib
>>> directories are not always located in the same folder.
>>>
>>
>> Yes, re-reading my own suggestion, you're right. Taking VTK as an
>> example, I
>> have here
>>
>> /usr/include/vtk
>> /usr/lib/vtk-5.0
>>
>> so yes, it is a bit more complicated. I'll try and think of another
>> solution.
>>
>>> Here is the answer: SWIG_generator.py uses GCCXML to parse OpenCascade
>>> headers and check all classes/methods to wrap. GCCXML performs a
>>> *recursive* parse of all headers. When processing the TopOpeBRep_*
>>> headers, I noticed that GCCXML didn't find the file
>>> TopOpeBRepDS_tools.hxx and the process fails. The
>>> 'TopOpeBRepDS_tools.hxx' is simply missing from the /inc OCC directory.
>>> I think it's a bug of OpenCascade but I didn't send any feedback to the
>>> Opencascade team. I just 'added' an empty TopOpeBRepDS_tools.hxx to the
>>> /inc directory so that GCCXML parsing is ok (note that SWIG does not
>>> need this file).
>>
>> Thanks for the explanation - I'm still trying to learn exactly how
>> you've
>> managed to generate these wrappers ;-)
>>
>>> The problem is that it requires, on Linux, root privileges since it
>>> modifies a directory that may have read only access for common user.
>>> It's a thing I didn't think about when hacking on Windows. Another
>>> soultion (Linux compliant) would be to:
>>> - create the missing header TopOpeBRepDS_tools.hxx in the
>>> SWIG_modular_linux_darwin directory
>>> - add this directory to the list of include paths so that g++ find the
>>> header.
>>> We could then remove all that weird lines from setup.py
>>
>> I like that solution, let's do that.
>>
>> Arthur
>>
>> _______________________________________________
>> Pythonocc-users mailing list
>> Pythonocc-users@gna.org
>> https://mail.gna.org/listinfo/pythonocc-users
>>
>
>
_______________________________________________
Pythonocc-users mailing list
Pythonocc-users@gna.org
https://mail.gna.org/listinfo/pythonocc-users
