notes on compilation Python with MinGW gcc
Ames Andreas (MPA/DF)
Andreas.Ames at tenovis.com
Mon Dec 1 15:33:47 EST 2003
More information about the Python-list mailing list
Mon Dec 1 15:33:47 EST 2003
- Previous message (by thread): notes on compilation Python with MinGW gcc
- Next message (by thread): notes on compilation Python with MinGW gcc
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hi Oleg, usenet at xmlhack.ru (Oleg A. Paraschenko) writes: > recently I compiled Python with MinGW gcc under Windows. > If there are some interest, you can look at log of process at > > http://uucode.com/texts/python-mingw/python-mingw.html > > In short, about compiling with MinGW gcc: > > * it is possible; > * result works; > * you must not do it unless you know for sure that you must do it; > * you must have a good knowledge of C and of debugging; > * in most cases all you need is to compile an extension module. I've recently taken a slitely different way to accomplish a native mingw32 build. You can get a patch from https://sourceforge.net/tracker/?func=detail&aid=841454&group_id=5470&at id=305470 As I haven't used msys yet and my real goal was to build a mingw32-python on linux (without the adventures of installing VC6 under wine, which reportedly works either). Thus it had to be a crosscompile, but you should be able to use cygwin (but you must edit xbuild-py.sh to do so and you must --build and --host on cygwin too, I've started the crossbuild on cygwin). The build still has quite a few 'rough edges' ;-): 1) The resource files (*.rc) are not yet compiled/linked in. Given that they are rather simple I wouldn't expect significant difficulties using windres. There would need to happen some changes to Makefile.pre.in though. 2) pythonw.exe is not yet built. Similar changes as above needed. 3) I couldn't yet convince scons that it would be a good idea to have a build directory seperate from the source tree. I have tried both the BuildDir function and the build_dir parameter to the SConscrript function but to no avail. It could either be a bug in scons or my own inability. As this was my first scons-script, I guess it's the latter. Some scons knowledgeable person might be able to fix this within seconds. Before you get a useful environment you must copy the Lib/ subdirectory from source- to the buildtree as well as all the *.pyd files from <source>/Modules and <source>/PC to <build>/DLLs or something. Note that running the testcases from a samba-share isn't good because the tests use filenames with an '@' sign which samba doesn't seem to like. You might be better off copying the complete buildtree to a native windows partition (if you build from linux anyway). The patch applies to 2.3.2 because I have no easy way to access the cvs repository. If there was interest in a native mingw32-python I would like to help to solve the above issues (and others I haven't found yet). But it doesn't appear that there is much interest in this, so I'd guess you can take the above as a toy project. cheers, andreas
- Previous message (by thread): notes on compilation Python with MinGW gcc
- Next message (by thread): notes on compilation Python with MinGW gcc
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Python-list mailing list