On 2014-10-01 15:32:47, Andreas Tille wrote: > Hi Antoine, > > On Wed, Oct 01, 2014 at 09:35:39AM -0400, Antoine Beaupré wrote: >> > I admit the request for sponsoring was sooooo long ago that I do not >> > remember and I see no value to think about aged code. I'd perfectly >> > agree if the current source for 3.2.2 would be used for packaging. I'd >> > recommend to use Files-Excluded if any files need to be stripped from >> > this source tarball (but I did not inspected it regarding this issue). >> >> I'm not familiar with this mechanism - but it certainly looks >> interesting! >> >> https://wiki.debian.org/UscanEnhancements >> >> Am I correct in understanding this would only simplify the >> get-orig-source target, not replace it? > > Yes. I typically use in d/rules > > get-orig-source: > uscan --verbose --force-download --repack --comress xz > > and list all files (with wildcards) in d/copyright under > > Files-Excluded > > This saves you the work of creating all times the same code just to > remove files from upstream source tarball. This will not recreate the same tarball reproducible. Unless there's something I don't know well about xzip / tar. >> > another look. Since you obviously had a more recent look and you do >> > not need a sponsor I'd trust your insight if you say it is OK. >> >> I don't think it's okay - it doesn't even build at that stage, >> especially because we strip so much off the tarball. > > I'd only strip files which are considered non-free (and perhaps some > files that might be only useful at non-Linux/non-BSD systems). Even then, it's a lot of files. >> > I personally did not. The only thing I could say that it always heats >> > my temper a bit if I learn about another instance of failed >> > communication between people working on free GIS software. I wonder >> > why we are not able to catch all those people into our common project >> > and do not reinvent the wheel over and over. :-( >> >> The PPA is from upstream, and is fairly minimal: it doesn't split the >> code in multiple packages (though I'm not sure why *we* do that in the >> first place) and doesn't attempt to deduplicate code. > > I do not see any reason why upstream should not commit packaging code to > git.debian.org and we check and upload this code. This is what we > established for several programs in the Debian Med team. Well, the reality is that upstream did very little packaging work, and probably have a wildly divergent opinion of what is acceptable in the Debian package. For example, their Debian package just ships a bunch of unsourced datafiles. >> > to Git or create a new Git repository which is compliant to Debian GIS >> > policy[1] (fetch the tarball via uscan and use >> > git import-orig --pristine-tar >> > to inject the source. >> >> Ugh.. pristine-tar... Why do we need this if we're going to strip out >> half the tarball anyways? > > I'm talking about the resulting tarball you get with the uscan above. > Specifically when using stripped (non-original with same MD5SUM as you > can download) it makes sense that all packagers can fetch easily a > (byte) identical source tarball which is only possible with > pristine-tar. I see no need to inject the downloadable tarball if > things wil lbe removed anyway. the get-orig-source should be doing this reproducibly, I think. It's what the current target I wrote does. a. -- Information is not knowledge. Knowledge is not wisdom. Wisdom is not truth. Truth is not beauty. Beauty is not love. Love is not music. Music is the best. - Frank Zappa
Attachment:
pgpho8oJ1wBxj.pgp
Description: PGP signature