A Friday 17 October 2008, escriguéreu: > Francesc Alted wrote: > | A Friday 17 October 2008, sk escrigué: > || Francesc Alted wrote: > ||| A Thursday 16 October 2008, sk escrigué: > |||||| I am trying to bundle an application using PyTables 2.0.4 with > |||||| PyInstaller. Running the bundled executable (on Windows XP if > |||||| it matters -- didn't try Linux yet) I get a traceback ending > |||||| with: > |||||| > |||||| File "hdf5Extension.pyx", line 11, in hdf5Extension > |||||| File "C:\pyinstaller\iu.py", line 427, in importHook > |||||| raise ImportError, "No module named %s" % fqname > |||||| ImportError: No module named utilsExtension > |||||| > |||||| In hdf5Extension.pyx I see this line: > |||||| > |||||| from utilsExtension cimport malloc_dims, get_native_type > |||||| > |||||| and wonder if it should be: > |||||| > |||||| from tables.utilsExtension cimport malloc_dims, > |||||| get_native_type > |||||| > |||||| but I don't know anything about pyrex. > ||||| > ||||| From the error, I'd say that you are using Python 2.6 for doing > ||||| the build. If this is the case, you are out of luck with the > ||||| 2.0.x series as they only support python 2.4 and 2.5. If you > ||||| want to use python 2.6, you may want to try packaging the trunk > ||||| version, accessible in: > ||||| > ||||| http://pytables.org/svn/pytables/trunk > ||||| > ||||| I'm in the process of releasing RC1 for 2.1 in trunk (I plan a > ||||| final 2.1 version for the end of the month), so this should be > ||||| fairly stable now. > ||||| > ||||| Cheers, > ||||| > ||||| -- > ||||| Francesc Alted > |||| > |||| No, this is on Python 2.5 (although I'm glad to hear about the > |||| 2.6 support). Is there a known PyInstaller bundling method for > |||| apps using PyTables? This is a big problem for our project. > ||| > ||| Mmh, can you try with the PyTables code in trunk? Chances are > ||| that the changes made for Python 2.6 would serve to fix the > ||| issue. > ||| > ||| Cheers, > || > || Francesc, thanks. Unfortunately installing from trunk on Windows > || is not easy. I got as far as installing HDF and pyrex and then > || easy_install won't fly because of the VC2003 issue and I don't > || know how to pass -c mingw32 thru easy_install. So I am trying > || setup.py now but, naturally, version.py chokes on the nonstandard > || MinGW ld version number (2.18.50.20080109). > | > | Which VC2003 issue? If it is a problem with the time the optimizer > | takes for compiling the numexpr extension (15 minutes or so on my > | laptop), you may want to try to relax the optimization level > | for this. > | Something like: > | > | python setup.py build_ext --cflags=/Ox > | > | [Incidentally, I don't remember well the correct optimization flag, > | just check the options of the VC2003 compiler] > | > | I normally use VC2003 to generate the binaries for Windows, so you > | should be able to do that too. > | > | Cheers, > > The "VC2003 issue" is that a) I don't have it, b) you can't get it, > and c) I don't want it (could screw up my VC2008 setup).
Yeah. All three good points. > I wanted to > use the VC2003 compatible MinGW GCC but that hit the ssize_t > definition problem. Yes. I've tried MinGW some time ago and I remember to hit this too. I think I've reported this to the HDF5 crew, and I don't remember exactly their answer, but I think that they don't support MinGW completely (however, they do support cygwin, if I recall correctly). > In any event, I got around the problem by making copies of the 2 pyd > files, as I indicated in my other message, and I don't have time to > keep fighting with the Windows build process to see if the trunk code > fixes it. But when you get to an RC with a binary installer I'd be > happy to try it with PyInstaller and report back. Hopefully I'll made a binary available for PyTables 2.1 and Python 2.5 by the end of the month. > Thanks for the help anyway -- and for PyTables. Glad that you like it. And thanks for reporting: I'm very interested in knowing about other ways to compile PyTables on Windows than using MSVC (I'm a Unix guy, and it is not always a trivial task to understand Windows compiler issues). I hope that, sooner or later, we would be able to use MinGW in order to compile PyTables. -- Francesc Alted ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Pytables-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/pytables-users
