Hi Joerg, I have resolved the problems. The probles related to version 2.6.1 only reside in python module itself and pyuno/loader (changing extension to .pyd). Other problems also exist in current version and I will handle them in an mingw port issue. I will raise another issue for Python 2.6.1 dependent to your original issue.
Best Regards, In message "Re: [porting-dev] python26 - looking for porters (issue 92413)", Takashi Ono wrote... >Hello Joerg, > >I am doing mingw port. Currently mingw port is still experimental and >actually python >module is not well tested. I want to make it really work. > >I am facing a couple of problems. > >1. Compilation of scripts >All the installed scripts are compiled when "make install" is invoked. >However, some >compilation fails as they include unicode escape '\\N'. The compiler tries to >load >unicodedata extension module but it is not yet installed. > >There is a workaround to add sharedinstall to the dependency rule of >libinstall and >append the directory to PYTHONPATH at the line invoing compileall.py, but I >do not know >why it does not occur on other platforms and am not sure it is a good fix. Do >they know >the location of the unicodedata dynamic loading module in temp directory? > >2. Stripping of symbols in python.exe >The symbols in binary files created in mingw build should be stripped when >making >install set. It is handled by a perl script >solenv/bin/modules/installer/strip.pm. Most >part of the code is copied from the one for unix platforms. It copies the >file to a >working directory, strip the symbols and save the name of the stripped file >to use >instead of the original source afterwards. >The problem is that the script uses only one directory for one language. The >files not >language specific use the same directory. The wrapper pyuno/python.exe for >windows is >overwritten by the python.exe in python-core during the stripping process. > >The current workaround is to handle pyuno/python,exe as an exceptional case >and use a >unique directory. I do not think it is an elegant solution. Maybe it is >better to >rename the core binary to have a unique name. I have not tried yet. > >3. QA-3 is failing >That is it. I would like to have suggestions where should I investigate. > >Best Regards, ---- Takashi Ono(HK Freak) mailto:[email protected] or [email protected] (Personal Address, checked every morning/evening and holidays) mailto:[email protected] (Address for business, checked every working days) http://www.hkfreak.net --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
