[02:18] <persia> Between the wiki and the /topic, I'm a bit confused.  For requesting sponsoring for a universe package, is the process of "attach a debdiff and subscribe the sponsors" still acceptable?
[02:24] <Unit193> I hope so, I've either been doing that or linking a dsc.
[02:29] <Unit193> persia: That is, yes subscribing the sponsors is how one gets it to show up on the queue.
[02:30] <persia> Cool.  Some things change, some don't.  Now to wait for a better time of day (as I recall this being about the time of day that #ubuntu-* went silent for a few hours) :)
[02:31] <persia> I'm surprised linking a .dsc works.  We used to get really fussy about that, as there was no way to demonstrate provenance (vs. having all the artifacts (or artifacts to generate the artifacts, like interdiffs) in launchpad).
[02:32]  * persia has read that interdiffs are now deprecated, to little surprise, as nobody else ever seemed to like them very much
[02:37] <Unit193> Well sometimes you have to, debdiffs have caused a need to re-upload before that I know of even.  The "preferred" way to do it is from a PPA build (see my irssi sponsor bug), but that allows you to upload a prospective version only once, otherwise you have to incriment the version (and sometimes sponsors will forget to change it back.)
[02:37] <Unit193> I have been known to just link from /source/ on my httpd. >_>
[02:38] <persia> Now I'm confused.  if working off the same tarball, I can't think of any reason a debdiff wouldn't be sufficient.
[02:39] <Unit193> Native package, png/jpg file changes.
[02:39] <persia> That said, I had a sponsor upload a debdiff from "persia@localhost" once, which remains an embarassment (and caused me, when I was a sponsor, to be extra careful checking changelog entries).
[02:39] <Unit193> Eg, xubuntu-docs. :P
[02:39] <Unit193> lintian would shout so very loudly now.
[02:39] <persia> Yes :)
[02:40] <Unit193> ...I presume all sponsors run that, at least.  It's part of my build process (pbuilder hook, after complete.)
[02:40] <persia> For generating binary files, I used to store them in 7-bit safe encodings (e.g. base64, uuencode, etc.), and then "build" them at package build time.
[02:40] <persia> When I was a sponsor, I used to run a bunch of quality checkers, and I think running lintian is part of the default calls to all the build tools these days.
[02:41] <persia> But I would be unsurprised if there was this one person who had a special workflow.
[02:42] <persia> Other way to do binary is to put the source under SCM, and then always get sponsored by a member of the team with SCM access.  For those few native packages that do branding, etc., (e.g. flavours), in some ways that makes more sense, as random folk not working in the flavour context probably shouldn't upload.
[02:43] <Unit193> Well at the time, neither one of us devs had upload rights, so it was a bit more fun.
[02:44] <persia> Yeah.  Nobody on the flavour team being an uploader is an issue.  I remember that being true intermittently for ubuntustudio for a while.
[02:46] <Unit193> (Both of us have packageset now, so it's all fine.)
[22:26] <Unit193> tsimonq2: Can you snag https://packages.qa.debian.org/v/variety/news/20180315T122744Z.html
[22:28] <tsimonq2> Unit193: .
[22:28] <tsimonq2> Unit193: Gah, forgot to -s unit193
[22:28] <tsimonq2> Unit193: Oh well.
[22:29] <tsimonq2> Unit193: When will you be able to do this yourself again? :P
[22:30] <Unit193> Well can't self approve, sooo...  Thanks!