2009/9/18 Simon Loic <simon1l...@gmail.com> > Hi, >
Hi Simon, > I'm new to PythonOcc, and found it very promising. Thanks for the work. > Welcome on board! > However it took me more than 3 days to make it compile and I'm not finished > yet. > Ouch. You should have ask a few questions on this forum earlier, this delay would have been shortened to 1 day max. I guess the following lines of your message are related to the compilation of both OCC/GEOM/pythonOCC on Ubuntu 9.04 64 bits. It's actually an excellent news: I've been trying to do the same under MacOSX Snow Leopard 64 bits (I recently moved from my old Dell/Windows laptop from a new MacBook Pro). Untill the 0.3 release, pythonOCC was developed with 32 bits systems, whether they are Windows or Linux based. The 64 bits compilation require a few hacks over the current pythonOCC code. I finally got OCC/pythonOCC running with 64 bits support, but the code is currently only available on the svn repository. The environment.py and setup.py scripts were modified to enable 64 bits support, but I still miss a few feedbacks from Win64 and Lin64 users. > Still I would like to share my experience to make it smoother for new users > and also for myself to get useful info. > You may think I am did a lot of crap so that it takes me 3 days!!! Well > this the story: > I took support on the wiki page: > http://www.pythonocc.org/wiki/index.php/Installing_pythonOCC_on_Linux > and checked out the http://svn.gna.org/svn/pythonocc/tags/0.3 revision of > pythonOcc. > > - As explained, I had to install OCC (1 day of hard struggle until I found > the pythonOcc wiki page > http://www.pythonocc.org/wiki/index.php/Installing_OpenCASCADE_on_Linux!<http://www.pythonocc.org/wiki/index.php/Installing_OpenCASCADE_on_Linux%21> > !) > This step normally goes without any problem. - Then a first missing (or not very clear) info is that you need to have the > Geom module of Salome installed in order for the compilation to work > (otherwise you have to specify -NO_GEOM and you get less functionalities) > - I thought that as only the GEOM module was required I would build it from > source : Such a huge mistake!!! their building scripts are based on > automake&co and need tons of environment variable to be specified so as for > the 3rd party libs to be found!! Actually I didn't make it work, until I > discovered that they had a hidden script allowing to use CMake. Afterwards > everything was just pleasure. (this step took me almost 2 days. > NO! It's certainly my fault since you took the wrong way, but you MUST NOT use the GEOM module from Salomé. Salomé is a real mess, and pythonOCC is based upon the GEOM module extracted from Salomé (and then independant from it). You MUST download/install rev.175 of the salomegeometry project ( http://sf.net/projects/salomegeometry). It's very easy to get it compile under Linux (a few minutes). I will upload to the wiki a simple Howto (I realize it's missing, sorry). > - Besides, another info was missing in the wiki : one have to set the env > variable SALOME_GEOM_LIB to the path where to find the GEOM libraries. > - That's not all, some of the Geom library names seems to have change > (that's my own explanation for what follows) as I had to change in setup.py > Sketcher by GEOMSketcher > GEOMImpl by GEOMimpl (downcase the i) > Archimede by GEOMArchimede > - Also, the fact that some headers have been renamed SGEOM_yyy.hxx is a > problem (i know it was done because of some windows conflicting behaviour) > because some of the functionalities have become inline (declaration is in > the header) in the last sources of Salome (5.1.2). So you get some undefined > reference to XXX::YYY(). As a consequence I had to regenerate cpp wrap files > (-generate_swig). And actually it's still running, I'm not sure it will > succeed) > Same thing: rather user salomegeometry. > - Finally, there is a small problem with the environment.py file : > platform.machine() returns x86_64 which is invalid for g++ -march flag ( > expects "x86-64 apparently). > Ok, thanks. > > I've attached a file patch.txt, but it's rather a summary of the workaround > I had to use than a real fix. Maybe some of these problems have been solved > in the trunk, or are just due to a misunderstanding from me. In this case I > hope some folks will correct the concerned points. And of course I wish I > can make it compile finally!!! > Cheers, > > Regards, > Loïc > Thomas
_______________________________________________ Pythonocc-users mailing list Pythonocc-users@gna.org https://mail.gna.org/listinfo/pythonocc-users
