Michael Biebl: > Hi Florian, hi release team > > On Sun, 26 Feb 2017 17:00:22 +0100 Florian Schlichting <f...@debian.org> > wrote: >> Control: tags -1 +patch >> >> Hi Michael, Berlin BSP here. >> >> Given that it's too late now to get a mallard-rng package into Stretch, >> I suggest to ship the mallard-1.0.rng file as part of the yelp-tools >> package for now (e.g. as /usr/share/yelp-tools/mallard/mallard-1.0.rng) >> and simply use that as relaxng schema in yelp-check: > > [..] > >> Do you want me to prepare an NMU or would you prefer to validate or >> improve upon the fix in some way? > > First of all, thanks a lot for working on this. > I've CCed the release team for their input. > > The issue is, that "yelp-check validate" requires network access as it > needs the mallard-1.0.rng schema which is not packaged yet [0]. >
Hi, Sorry, this email fell through on our list. > Now, according to [1], entangle is the only source package actually > running "yelp-check validate" during build [1]. If there is no network > access this doesn't lead to a failed build though, as "yelp-check > validate" always returns 0, even on error (which I consider to be a bug > [2], fwiw). > > Shipping a copy of the mallard-rng schema in yelp-tools is a workaround > and I guess Florian agress that packaging the mallard-rng schema is what > should be done (and help with that would be grealy appreciated). > > I'm undecided whether we should apply the workaround for stretch given > that we don't produce any build failures, that's why I'd like some input > from the release team on this: > > Should we apply Florian's patch [3] or tag the bug as stretch-ignore? > Other suggestions? > > Regards, > Michael > > > > [0] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=788769 > [1] https://codesearch.debian.net/search?q=yelp-check+validate > [2] https://bugzilla.gnome.org/show_bug.cgi?id=779615 > My take: 1) Shipping a the schema embedded in yelp-tools is ok with me for stretch (just remember to update d/copyright as necessary) 2) Re: "yelp-check validate" - I agree it /ought/ to fail on error. Assuming that is what upstream intended, I am fine with a patch for that. I would prefer that at least 1) was applied for stretch as it seems rather trivial and better than the status quo. Thanks, ~Niels
signature.asc
Description: OpenPGP digital signature
