severity 674649 serious kthxbye On Sat, May 26, 2012 at 04:47:07PM +0200, Philipp Schafft wrote: > reflum, > > On Sat, 2012-05-26 at 21:37 +0930, Ron wrote: > > As you know, we're planning on removing the celt package from Wheezy. > > Please disable celt support at the soonest opportunity so that we can > > move ahead with doing that before the freeze. This codec is obsolete > > now and should no longer be used by general purpose applications. > > CELT adds a lot pain to maintainers and upstream. So I'm with you that > it should be removed.
You're missing the point. CELT isn't a pain and almost never has been, except to people who didn't pay attention. It was an experiment. Now the experiment is over and so it is obsolete, and henceforth unmaintained. > Yet removal requires a good transtion. The first part was Opus entering > Debian. You're still missing the point. Opus releasing just means celt really is dead and never will be resurrected. There is no 'transition', they aren't compatible. So the first and only part here is removing celt. If you want to add support for Opus now or later, that is an entirely separate question. But it's not a prerequisite for removing celt now. > As of my understanding this was slowed down by legal problems. Your understanding is wrong. You've had explicit permission to develop and test with this for over 2 years now. You've been lurking in places where everybody who would listen was begged repeatedly to develop and test with it and report any bugs that they found. The only thing that we were asked not to do was distribute it for general purpose use or in commercial products until the bitstream was frozen and the working group had signed off on a final version of it. > The same problems slowing down (src:roaraudio) upstream development. That 'problem' is entirely of your own invention. You've been nagging me about when this will be ready for over a year - at no point in that did you actually mention that you've done *nothing* in all this time. It's a bit late to find and report bugs against a spec after it is frozen. But boggling about that is unrelated to this bug. > The next step is to implement the needed support for Opus in upstream. > See http://bts.keep-cool.org/ticket/243 for details. This is unrelated to this bug. > This takes about two upstream release cycles (~2*1 month). Of cause you > can speed this up by helping the upstream team with implementing the > needed support. This is unrelated to this bug. > Last step is to give users the time to update their config and convert > all files encoded in CELT (this may also apply to other packages > providing file based operations). This is ... wait, What?? You've been encouraging people to create *files* encoded with CELT? Srsly?? Let me get this straight. You added support for an experimental codec, that was designed specifically for real-time, low-latency streaming and not at all for archival purposes. That has never had 2 releases which are bitstream compatible. And then ... you told your users they could store their data in files encoded with it ... ? What did you think they would read them with after the next CELT version was uploaded? I'm making this bug release critical on the basis of that. This is data loss waiting to happen, if it hasn't happened already. There is no way that Debian should ship with something that gets people to put their data into files they will never be able to read ever again. > I already wrote a mail to the RA/RS-Announce list to speed up this last > step. Still people can not yet switch over because this depends on the > not yet implemented support. I would guess this will need one or two > months itself. > > So I consider this transition doable in about 4 months. As you are very, > very late in the Debian release cycle I don't think this can be done > before the upcoming freeze. I'm pretty sure I can upload a version that disables CELT in about 15 minutes. Which should give you plenty of time to do that yourself before I consider doing so myself. If you don't have time to implement Opus support for this release, then that is fine. These two things are not directly related. And this is not even close to a reason to continue shipping a codec experiment that is obsolete and now unsupported -- in fact you've just become the poster child for precisely why we should not ... You've been lurking around the people who developed this for longer than I can remember now - and if you screwed this up so badly that you had people encoding files with it, and had no idea that its future ended with the IETF last call for Opus - then there is no way that we should continue to ship this for a moment longer than is needed to remove it, lest other users, who aren't so in the loop, also make similar mistakes regarding it. So please, stop making unrelated excuses and just disable celt now. You aren't doing anybody, let alone your own users, any favours with that. It sounds like this is one application that should have never ever enabled it for public release in the first place. Thank you. Ron -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
