[Python-modules-team] Bug#911904: Processed: block 911904 with 911915

[073plan at gmail.com]

Hi all,

在 2018-10-26五的 13:49 +0200，Michal Čihař写道：
> Hi
> 
> On Fri, 2018-10-26 at 10:51 +0200, Ondřej Surý wrote:
> > LIBTIDY_LIBRARY=$(shell readlink -f
> > /usr/lib/$(DEB_HOST_MULTIARCH)/libtidy.so)
> 
> Well it all started from this file being absent from the package :-)
> 
> > LIBTIDY_PACKAGE=$(shell dpkg-query --search $(LIBTIDY_LIBRARY) | cut
> > -f 1 -d :)
> > LIBTIDY_LIBRARY_FILE$(shell basename $(LIBTIDY_LIBRARY))
> > 
> > then
> > 
> > sed -e “s/libtidy.so/$(LIBTIDY_LIBRARY_FILE)” lib.py
> > 
> > echo libtidy:Depends=$(LIBTIDY_PACKAGE) > python-utidylib.substvars
> > 
> > and replace
> > 
> > libtidy5deb1 | libtidy5,
> 
> I don't get what this would solve - there is nothing wrong with
> alternate dependecies as long as they both provide libtidy.so. Anyway
> same is in Build-Depends as well, where substvars are not really
> helpful.

So what will be the final solution? I'm wondering if we could solve the
problem soon to complete the transition.

BTW I see the advantage of automatically generating lib dependency; however, I
also believe that listing alternate dependencies should be okay as well since
it extends the compatibility range of binary package. Manual adjustment will
be necessary in each transition but as long as the package maintainer is
comfortable with that, it should be acceptable.


--
Regards,
Boyuan Yang