On Mon, 2007-05-14 at 17:03 +0200, Lucas Nussbaum wrote: > On 14/05/07 at 09:58 -0400, Adam C Powell IV wrote: > > severity 423794 normal > > thanks > > > > On Mon, 2007-05-14 at 11:09 +0200, Jérémy Bobbio wrote: > > > Package: pysparse > > > Version: 1.0-1 > > > Severity: serious > > > Justification: no longer builds from source > > > > > > Hi! > > > > > > pysparse currently fails to build from source in sid [1]: > > > > > > --- 8< --- > > > running build_ext > > > building 'pysparse.spmatrix' extension > > > creating build/temp.linux-i686-2.5 > > > creating build/temp.linux-i686-2.5/Src > > > gcc -pthread -fno-strict-aliasing -DNDEBUG -g -O2 -Wall > > > -Wstrict-prototypes -fPIC -IInclude -I/usr/include/python2.5 -c > > > Src/spmatrixmodule.c -o build/temp.linux-i686-2.5/Src/spmatrixmodule.o > > > Src/spmatrixmodule.c:13:37: error: Numeric/arrayobject.h: No such file or > > > directory > > > In file included from Src/spmatrixmodule.c:24: > > > Src/ll_mat.c: In function 'LLMat_matvec_transp': > > > Src/ll_mat.c:763: error: 'PyArrayObject' undeclared (first use in this > > > function) > > > Src/ll_mat.c:763: error: (Each undeclared identifier is reported only once > > > [...] > > > error: command 'gcc' failed with exit status 1 > > > make: *** [stamp-build-2.5] Error 1 > > > --- >8 --- > > > > That's odd. Sounds like something major changed in python-numeric, > > since its header file is missing... no, unstable still has 24.2-7 like > > etch. > > Yes, but 24.2-7 doesn't include the header file for python 2.5.
Right, see below. > > Have you tried rebuilding python-numeric with python 2.5, then building > > pysparse? I notice that python-numeric does not appear in > > http://people.debian.org/~lucas/logs/2007/05/13/ , nor in any other > > subdir of http://people.debian.org/~lucas/logs/2007/ , so it seems not. > > if it's not here, it just means that it never failed. How do we know it was ever tried? > > I see. This is a problem with the new python build system: it's hard to > > tell which packages have been built with which version of python. > > There's no reason for me to change my python-numeric dependency, but no > > way for any build system to tell whether the same python-numeric package > > has been built for the new python. > > > > Reclassifying this as normal. If you can show that you built its python > > dependencies with 2.5, and then this failed, then I will agree that it > > is a bug in pysparse. > > I'm not really a python expert. :-) But why did your package suddenly > start to build python 2.5 files as well ? When you try to build a python package, it attempts to build all of the current default versions. Last Fall when I first uploaded pysparse, it automatically built and installed headers, .pyc files, etc. for python 2.3 and 2.4. Now there is python 2.5, so when you try to build pysparse again, it tries to build python 2.5 files. But python-numeric was built before python 2.5, so it only built headers, pyc etc for 2.3 and 2.4 (maybe not even 2.3, I'm not sure). So in order to have python-numeric headers for python 2.5, you need to rebuild that package now with python 2.5. Otherwise, when rebuilding pysparse, it will try to build 2.5, and look for python-numeric 2.5 header files, but they won't be there! So please try building python-numeric, install the newly-built version, then try building pysparse. As I mentioned, this is a general problem with python upgrades using the new python build system (introduced about a year ago IIRC). After a new python version enters the defaults list, attempting to rebuild an arbitrary python package is likely to fail, unless all of its dependencies have been rebuilt and uploaded first. Please verify that it builds as I described above (when you rebuild and install python-numeric first), then we can close this bug. Cheers, -Adam -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Welcome to the best software in the world today cafe! http://www.take6.com/albums/greatesthits.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
