I added back scm-migration-dev to the cc list, for the Python compilation question.
>> So the .py.po rule is my only question, mainly because I haven't used >> Python for anything with localization. > > Deltas on last delta (on top of "webrev-mjn"), showing removed .py.po: > > http://zhadum.east/ws/carlsonj/perl-rule/webrev-mjn-2/ > > Updated full webrev: > > http://zhadum.east/ws/carlsonj/perl-rule/webrev/ > > &^%& Python. :-/ Indeed. I think I'm OK with this, in that it works (I believe correctly) for a script that is meant to be invoked (using python as an interpreter) as a top level program. For something that would be imported as a module by another Python program, we probably also need a .py.pyc suffix rule, but that's not the bug you're fixing now. Sadly, "541 # Rule for Python (no i18n supported yet)" makes it sound like Python doesn't support i18n, instead of "we don't support extracting messages from Python code." That might actually be correct, because the gettext.py folks seem to have completely punted on using portable message objects (.po), and on the Solaris message catalog format (.mo), and gone straight to GNU-only message objects (.mo). You know, why use existing tools like xgettext and msgfmt, when you can reinvent them instead? :( For Jim: I'm done ranting now. Your changes look good. For scm-migration-dev folks: do we need a .py.pyc suffix rule? --Mark
