On Sat, September 17, 2016 12:09 pm, Daniel Jakots wrote:
> On Sat, 17 Sep 2016 11:28:29 -0400, trondd <tro...@kagu-tsuchi.com>
> Any reason to remove this chunk? It doesn't package right now:
> $ make package
> `/usr/ports/pobj/gpodder-3.9.1/fake-amd64/.fake_done' is up to date.
> ===> Building package for gpodder-3.9.1
> Create /usr/ports/packages/amd64/all/gpodder-3.9.1.tgz
> | 59%
> Error: /usr/ports/pobj/gpodder-3.9.1/fake-amd64/usr/local/man/man1/gpo.1
> does not exist
> does not exist
> /usr/ports/pobj/gpodder-3.9.1/fake-amd64/usr/local/man/man1/gpodder.1 does
> not exist
> Fatal error: can't continue
> so I looked for them and:
> /usr/ports/pobj/gpodder-3.9.1/fake-amd64$ find . -name gpo.1
> With the mv, it packages again.
My fault. My patch didn't include removal of patches/patch-setup_py which
breaks upstream's handling of the man paths.
> make lib-depends-check complains:
> Extra: iconv.6 intl.6
> so I removed devel/gettext modules and textproc/intltool. I don't know
> if it needs to be in WANTLIB but it doesn't complain anymore at least :)
Thanks, forgot to check this. intltool is a build dependency.
On that note, I'm confused when to use a *_DEPENDS and when to use
WANTLIB. In this case it calls the intltool executable, rather than
linking to a library so maybe that's the obvious answer.
> make test fails here:
> ===> Regression tests for gpodder-3.9.1
> LC_ALL=C PYTHONPATH=src/ /usr/local/bin/python2.7 -m gpodder.unittests
> No handlers could be found for logger "gpodder.util"
> Traceback (most recent call last):
> File "/usr/local/lib/python2.7/runpy.py", line 174, in
> "__main__", fname, loader, pkg_name)
> File "/usr/local/lib/python2.7/runpy.py", line 72, in _run_code
> exec code in run_globals
> line 89, in <module>
> cov = coverage.Coverage()
> AttributeError: 'module' object has no attribute 'Coverage'
> gmake: *** [makefile:64: unittest] Error 1
The problem here is that I didn't have py-coverage installed. They
require a newer version than we have in ports. I don't know if the newer
one would be backwords compatible or not. Can someone more experienced
with Python ports help out here? If I update py-coverage, what do I need
to check for fallout?