Comment 6 for bug 10768

Revision history for this message
In , Peter Moulder (peter-moulder) wrote : comments & suggested method of attack

As mentioned earlier, I'd start with touch src/ink{scape,view} && make
-k check and see what test programs do or don't link: different test
programs link against a different number of libraries.

The builder successfully built inkscape 0.39-7 on Oct 19, and failed to
build 0.40pre1 on Nov 5. First we should try to establish whether the
failure is due to a change in the build environment or to the change in
inkscape (e.g. due to the additional libraries inkscape 0.40pre1 links
against that 0.39-7 didn't).

I'm inclined to get the simplest thing out of the way first, so I'd
start with trying to build 0.39-7 again (which has previously compiled
successfully on mips{,el}). If it now fails, then this strongly
suggests that the bug is due to a change in the build environment.

However, my guess is that the problem is connected to one of the
dependencies added between 0.39 and 0.40. One way of finding which
would be a binary search through CVS versions, paying particular
attention to dates suggested by `cvs log configure.in'. (Most of the
new dependencies were added before configure.in was renamed to
configure.ac.) One thing to be careful of is that debian/control in CVS
may not have been kept in sync with the new dependencies, so assume the
Build-Depends of 0.40 if using a chroot environment.

pjrm.