On 13/12/16 23:15, Santiago Vila wrote: > On Tue, 4 Oct 2016, Santiago Vila wrote: > >> But in either case, look at the error message. It's essentially like >> this: >> >> /usr/bin/install -c -m 644 >> some-files _generated/errors.py some-files _generated/errors.py some-files >> directory >> >> This fails because _generated/errors.py is listed twice. > > And in fact the patch below shows that $(BUILT_SOURCES), which is what > the install target tries to install, contains _generated/errors.py twice. > > To test it, apply the patch, build the package with "dpkg-buildpackage -A" > and then try: > > make -C src show_BUILT_SOURCES | grep --color _generated/errors.py > > You should see _generated/errors.py twice, one because it's listed > explicitly and another one as the result of the patsubst in spec_files. > > The question, then, is not why it fails, but why it does not always fail. > > Please try to explain how the current Makefile tries to avoid the > failure. I guess it relies on wildcard $(spec_dir)/*.xml having > a particular order, but the order of such wildcard expansion > depends on filesystem ordering, so it may be anything. > > So I really think there is a Makefile bug here, even if can't > explain it in full. please forward the bug upstream.
This module is dead upstream. It is replaced by gir1.2-telepathy-glib-*. It won't see any upstream fixes at all. I want to get it removed from Debian, but there are a few rdeps. I'll probably do that anyway once stretch is released. Cheers, Emilio