Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 11:09, Diego 'Flameeyes' Pettenò wrote: On Friday 14 July 2006 16:43, Chris Gianelloni wrote: While it is a working solution, it isn't necessarily a sensible one. You can take over xine-lib and fix it however you prefer. As this, as well as any other idea you can find, is just an HACK until portage devs implements the per-package use.mask that i asked WAY before 2.1 release, but was then left OUT of the freeze and thus of the featureset we can use 6 months from now, I REFUSE to change the behaviour. /me humps angry over worked flameeyes -mike pgp8xnHpdzbVK.pgp Description: PGP signature
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 06:03, Daniel Watkins wrote: Is there a rationale behind this decision? On some systems libmad does not work and has to be masked, if I called it mp3, it couldn't be use.masked or all the mp3 supports, even when not provided by libmad, would have been removed. Per-package use.mask is not here for another year and in the mean time I needed a working solution, this is it. I would also say that it helps not to overload the same useflag with different meanings, as we already seen a couple of times where v4l and v4l2 useflags are better to be two different things, ibidem for qt3/qt4. -- Diego Flameeyes Pettenò - http://farragut.flameeyes.is-a-geek.org/ Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE pgpYqhpoelYus.pgp Description: PGP signature
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 2006-07-14 at 05:03 +0100, Daniel Watkins wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 It seems to me that having both of these two flags can only cause confusion. I upgraded xine-lib yesterday and spent a very frustrating 2 hours trying to work out what had broken my Amarok MP3 playback. It turns out that having the 'mp3' USE flag set globally is not good enough to get MP3 playback enabled in xine-lib, you need 'mad' set. Is there a rationale behind this decision? If not, it would seem to be quite an important issue, as a lot of users will be looking for MP3 playback (and expecting it to work from one version of xine-lib to the next, without having to play with USE flags). I think the problem is that the flags have started to become used for different things. It *used* to be like this: mp3 - enabled mp3 support mad - used libmad over $whatever for mp3 playback What this meant was simple. If a package *only* used libmad, then USE=mp3 would enable it. The *only* reason you would use USE=mad is if, for example, a package used libmpeg3 *or* libmad, to select between the two. Some people have started to interpret the this package uses libmad as you need USE=mad for mp3 support which, in my opinion, is wrong. If a user has USE=mp3 -mad then they should *always* have working mp3 support. -- Chris Gianelloni Release Engineering - Strategic Lead x86 Architecture Team Games - Developer Gentoo Linux signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 2006-07-14 at 16:20 +0200, Diego 'Flameeyes' Pettenò wrote: Per-package use.mask is not here for another year and in the mean time I needed a working solution, this is it. While it is a working solution, it isn't necessarily a sensible one. If the package only uses libmad, and libmad doesn't work on $platform, then why not simply disable mp3 support on that platform until libmad is fixed? Why *force* people to use USE=mp3 mad on a platform where it *does* work? I would also say that it helps not to overload the same useflag with different meanings, as we already seen a couple of times where v4l and v4l2 useflags are better to be two different things, ibidem for qt3/qt4. It isn't overloading it, at all. USE=mp3 means I want mp3 support and not I want mp3 support unless it happens to use libmad in which case I also need to add mad to my USE. Seriously, if the mp3 support via libmad doesn't work on a platform, drop it. Don't make everyone else jump through hoops because a few platforms don't support it. -- Chris Gianelloni Release Engineering - Strategic Lead x86 Architecture Team Games - Developer Gentoo Linux signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 16:38, Chris Gianelloni wrote: If a user has USE=mp3 -mad then they should *always* have working mp3 support. Give me per-package use.mask baby, and I'll do whatever you want. But as when I asked it was considered low priority, then you can start barking at portage devs, instead of me. -- Diego Flameeyes Pettenò - http://farragut.flameeyes.is-a-geek.org/ Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE pgpeSAxueMUb7.pgp Description: PGP signature
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 14 Jul 2006 10:43:54 -0400 Chris Gianelloni [EMAIL PROTECTED] wrote: | If the package only uses libmad, and libmad doesn't work on $platform, | then why not simply disable mp3 support on that platform until libmad | is fixed? Because it confuses the hell out of users who want to know why they're not getting mp3 support despite a package being shown as being built with USE=mp3. -- Ciaran McCreesh Mail: ciaran dot mccreesh at blueyonder.co.uk -- gentoo-dev@gentoo.org mailing list
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 16:43, Chris Gianelloni wrote: While it is a working solution, it isn't necessarily a sensible one. You can take over xine-lib and fix it however you prefer. As this, as well as any other idea you can find, is just an HACK until portage devs implements the per-package use.mask that i asked WAY before 2.1 release, but was then left OUT of the freeze and thus of the featureset we can use 6 months from now, I REFUSE to change the behaviour. You should know better than me this problem, considering that you masked media-video/transcode on x86 2.4 profile because it was depending on linux-headers 2.6 (as it should have been) with v4l2 useflag enabled. You should have use.masked (as it was later done) that useflag as it's 2.6 specific, and I did it that way to satisfy sparc requirement of not having v4l useflag present for them, as it failed with 2.4 kernel. So if I didn't do it that way, forcing people wanting v4l support to use v4l v4l2, now I should still have the stupid, broken, idiotic transcode 0.6 in the tree, or one of the x86 deptrees broken, or one of the x86 deptrees with NO v4l at all. -- Diego Flameeyes Pettenò - http://farragut.flameeyes.is-a-geek.org/ Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE pgpks64mzEi0Z.pgp Description: PGP signature
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Fri, 14 Jul 2006, Diego 'Flameeyes' Petten?? wrote: But as when I asked it was considered low priority, then you can start barking at portage devs, instead of me. I know some of us arch team ninjas asked for this in bug #96368 some time ago. For us it would be very very handy to be able to do this rather than lots of stuff like foo? (!arch bar baz) and the like as well as printing out additional ewarns to note that this use flag doesn't work on this arch for this package that no one ever sees. Cheers, - -- Jason Wever Gentoo/Sparc Team Co-Lead -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.4 (GNU/Linux) iD8DBQFEt8cSdKvgdVioq28RAiHIAKCOvHET5vWYJ1tL1boX16VKY/WrzgCgklpS SOmbeqoQVdyDuIoClI1FHYA= =T0Pq -END PGP SIGNATURE-
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 2006-07-14 at 15:59 +0100, Ciaran McCreesh wrote: On Fri, 14 Jul 2006 10:43:54 -0400 Chris Gianelloni [EMAIL PROTECTED] wrote: | If the package only uses libmad, and libmad doesn't work on $platform, | then why not simply disable mp3 support on that platform until libmad | is fixed? Because it confuses the hell out of users who want to know why they're not getting mp3 support despite a package being shown as being built with USE=mp3. Uhh... einfo/elog/ewarn/etc... -- Chris Gianelloni Release Engineering - Strategic Lead x86 Architecture Team Games - Developer Gentoo Linux signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 2006-07-14 at 17:09 +0200, Diego 'Flameeyes' Pettenò wrote: On Friday 14 July 2006 16:43, Chris Gianelloni wrote: While it is a working solution, it isn't necessarily a sensible one. You can take over xine-lib and fix it however you prefer. Quite honestly, I don't care. As this, as well as any other idea you can find, is just an HACK until portage devs implements the per-package use.mask that i asked WAY before 2.1 release, but was then left OUT of the freeze and thus of the featureset we can use 6 months from now, I REFUSE to change the behaviour. See, it is this kind of self-serving attitude that really needs to stop around here. So the portage devs didn't include something that you wanted in the latest release... Did you give them a patch for it? Making decisions like this that confuse our users and make the distribution harder to use and more inconsistent because you have a personal beef with another team simply because they didn't drop everything and do what you wanted them to work on instead of what they already had on their plates does not help our distribution. Instead, it makes us look like a bunch of selfish little babies. No thanks. I would much rather have a consistent and working distribution than cater to some childish self-important bullshit that only harms our users. You should know better than me this problem, considering that you masked media-video/transcode on x86 2.4 profile because it was depending on linux-headers 2.6 (as it should have been) with v4l2 useflag enabled. You should have use.masked (as it was later done) that useflag as it's 2.6 specific, and I did it that way to satisfy sparc requirement of not having v4l useflag present for them, as it failed with 2.4 kernel. So if I didn't do it that way, forcing people wanting v4l support to use v4l v4l2, now I should still have the stupid, broken, idiotic transcode 0.6 in the tree, or one of the x86 deptrees broken, or one of the x86 deptrees with NO v4l at all. I don't give a crap about transcode. I was talking about USE=mp3. Changing the subject doesn't help anyone. Anyway, I can see that I'm just going to be ignored, and since I don't have the time to personally take over xine-lib, I'm going to drop it and just let you do whatever it is you think is best instead. -- Chris Gianelloni Release Engineering - Strategic Lead x86 Architecture Team Games - Developer Gentoo Linux signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
Diego 'Flameeyes' Pettenò wrote: On Friday 14 July 2006 06:03, Daniel Watkins wrote: Is there a rationale behind this decision? On some systems libmad does not work and has to be masked, if I called it mp3, it couldn't be use.masked or all the mp3 supports, even when not provided by libmad, would have been removed. Per-package use.mask is not here for another year and in the mean time I needed a working solution, this is it. In the specific case of xine-lib, the mad USE flag can simply be replaced with mp3 because mips, which is the only arch that has the mad USE flag in use.mask, doesn't have any version keyworded. Of course that doesn't fix the general problem, but at least it would save time for a lot of users... -- Kind Regards, Simon Stelling Gentoo/AMD64 Developer -- gentoo-dev@gentoo.org mailing list
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 14 Jul 2006 14:14:39 -0400 Chris Gianelloni [EMAIL PROTECTED] wrote: | On Fri, 2006-07-14 at 15:59 +0100, Ciaran McCreesh wrote: | On Fri, 14 Jul 2006 10:43:54 -0400 Chris Gianelloni | [EMAIL PROTECTED] wrote: | | If the package only uses libmad, and libmad doesn't work on | | $platform, then why not simply disable mp3 support on that | | platform until libmad is fixed? | | Because it confuses the hell out of users who want to know why | they're not getting mp3 support despite a package being shown as | being built with USE=mp3. | | Uhh... einfo/elog/ewarn/etc... Are ignored. And even ignoring that, what're you going to do when $platform gets support for libmad on certain profiles? -- Ciaran McCreesh Mail: ciaran dot mccreesh at blueyonder.co.uk -- gentoo-dev@gentoo.org mailing list
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 20:20, Chris Gianelloni wrote: See, it is this kind of self-serving attitude that really needs to stop around here. So the portage devs didn't include something that you wanted in the latest release... Did you give them a patch for it? Do I ask people to give me patches to xine-lib when they report bugs? Making decisions like this that confuse our users and make the distribution harder to use and more inconsistent I find disputable that my solution is harder to use and more inconsistent. I actually find it more consistent across all the arches and platforms we have, as Ciaran already said. because you have a personal beef with another team simply because they didn't drop No I didn't do that decision because they didn't put per-package use.mask . I did that decision because the only other way to have the same behaviour is resorting to the arch hacking thing that is not acceptable to my eyes. Both solutions aren't beautiful, I still find my solution better than the arch hacking. What I'm saying with that thing about portage is that you cannot get to me screaming that I used an improper solution because the only proper solution was _postponed_ by portage team. Which means that if you really don't like this situation you have to tell them, not me. No thanks. I would much rather have a consistent and working distribution than cater to some childish self-important bullshit that only harms our users. Exactly my point, so why instead of starting criticising my choice in favour of your choice you don't go writing the famous patch? By the way, a patch there was, by antarus, but needed to be cleaned up. My python skills are weak, and you really don't want to see me hacking to portage, but KingTaco iirc offered to take a look. But freeze was entered before anybody could do anything. I don't give a crap about transcode. I was talking about USE=mp3. Changing the subject doesn't help anyone. I'm not changing the subject, I'm just showing you that xine-lib's USE=mad is just _one_ of many other similar problems that don't have a proper solution. -- Diego Flameeyes Pettenò - http://farragut.flameeyes.is-a-geek.org/ Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE pgpUUxX9TQR1D.pgp Description: PGP signature
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 20:24, Simon Stelling wrote: In the specific case of xine-lib, the mad USE flag can simply be replaced with mp3 because mips, which is the only arch that has the mad USE flag in use.mask, doesn't have any version keyworded. If I revert the change, mips won't be able to re-keyword xine-lib anytime soon. And I try to make sure that everything I maintain can be used as widely as possible. It would be even worse if I now change to mp3, and then mips ask to keyword and I have to change it back to mad. -- Diego Flameeyes Pettenò - http://farragut.flameeyes.is-a-geek.org/ Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE pgpDMuWZ0zwIC.pgp Description: PGP signature
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Fri, 2006-07-14 at 14:20 -0400, Chris Gianelloni wrote: On Fri, 2006-07-14 at 17:09 +0200, Diego 'Flameeyes' Pettenò wrote: On Friday 14 July 2006 16:43, Chris Gianelloni wrote: While it is a working solution, it isn't necessarily a sensible one. You can take over xine-lib and fix it however you prefer. I'd like to apologize for my last email. I sent it a bit too hastily and really didn't think. It definitely read as inflammatory against Diego, which was not my intention, as he does wonderful work. However, it still has some points about the general attitudes that are prevalent within Gentoo of not doing what is best for our users simply because someone else doesn't do what we want. We really need to stop and think about the users before we make changes like this. All of us. -- Chris Gianelloni Release Engineering - Strategic Lead x86 Architecture Team Games - Developer Gentoo Linux signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Diego 'Flameeyes' Pettenò wrote: Per-package use.mask is not here for another year and in the mean time I needed a working solution, this is it. It think we can have it sooner than another year. There are lots of fixes in 2.1.1_pre and I'd like to close the merge window pretty soon so that it can be stabilized. I'll work on a patch for package.use and package.use.mask so that we should be able to have it in a stable 2.1.1 release within a month or two. Zac -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.4 (GNU/Linux) iD8DBQFEt/11/ejvha5XGaMRAjR1AKDZ9lWLwjJ0BUdQv7HTg9RM1xzVbgCdHLhZ h4k/BfvdaytVeJ42Zyv3H1I= =WJ9n -END PGP SIGNATURE- -- gentoo-dev@gentoo.org mailing list
Re: [gentoo-dev] 'mad' vs 'mp3' USE flags
On Friday 14 July 2006 22:24, Zac Medico wrote: It think we can have it sooner than another year. There are lots of fixes in 2.1.1_pre and I'd like to close the merge window pretty soon so that it can be stabilized. I'll work on a patch for package.use and package.use.mask so that we should be able to have it in a stable 2.1.1 release within a month or two. That would be great, of course that goes a bit against your current schedule, no? :) -- Diego Flameeyes Pettenò - http://farragut.flameeyes.is-a-geek.org/ Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE pgpx4KFO61r1F.pgp Description: PGP signature