reopen 476976 found 476976 0.98.2-1 thanks Debian Bug Tracking System <[EMAIL PROTECTED]> (25/04/2008): > This is an automatic notification regarding your Bug report > which was filed against the scons package: > > #476976: scons: Doesn't work at all > > It has been closed by Mark Brown <[EMAIL PROTECTED]>.
| Changes: | scons (0.98.2-1) unstable; urgency=low | . | * New upstream release. | * Build against new python-central, picking up new generated code which | fixes upgrades (closes: #476976). Hmm, no, the upgrade isn't fixed. Context: -------- It started with my wondering whether I was hitting a regression in scons, which you can trigger by trying to build blender in a clean sid environment. AFAIUI by reading the diff, the Options->Variables transition is being initiated, being at the moment mostly transparent, then it's planned to warn for deprecation, and then to remove that compat layer completely. While I'm perfectly OK with the above plan, it looks like that compat layer just doesn't work. | scons: Reading SConscript files ... | ImportError: No module named BoolOption: | File "/home/kibi/hack/collab-maint/blender-2.45/SConstruct", line 44: | import tools.btools | File "/home/kibi/hack/collab-maint/blender-2.45/tools/btools.py", line 4: | import SCons.Options.BoolOption | make: *** [build-stamp] Error 2 How I reproduced #476976: ------------------------- Blender FTBFS with scons/0.98.2-1. Wondering which version was OK, I downgraded scons to its testing version (0.98.0-1), which doesn't have the Option->Variable transition, Blender built fine. Upgrading to the sid version of scons again, Blender built fine again. And when purging it: | $ sudo dpkg -P scons | (Reading database ... 26708 files and directories currently installed.) | Removing scons ... | dpkg - warning: while removing scons, directory `/usr/lib/scons/SCons' not empty so not removed. | dpkg - warning: while removing scons, directory `/usr/lib/scons' not empty so not removed. | [EMAIL PROTECTED]:~/hack/collab-maint/blender-2.45$ find /usr/lib/scons/ | /usr/lib/scons/ | /usr/lib/scons/SCons | /usr/lib/scons/SCons/Options | /usr/lib/scons/SCons/Options/EnumOption.pyc | /usr/lib/scons/SCons/Options/ListOption.pyc | /usr/lib/scons/SCons/Options/PathOption.pyc | /usr/lib/scons/SCons/Options/__init__.pyc | /usr/lib/scons/SCons/Options/BoolOption.pyc | /usr/lib/scons/SCons/Options/PackageOption.pyc All of this being in an uptodate i386 chroot; python-central is 0.6.5. Maybe you'll need to embed a code snippet in a maintainer script to clean the mess that previous python-central version(s) introduced/left in your previously-uploaded scons packages? Anyway, the above-mentioned regression deserves its own bugreport, which is on its way. Mraw, KiBi.
signature.asc
Description: Digital signature