| View previous topic :: View next topic |
| Which of variants in post 1 would you prefer? |
| A (USE=libav for libav, USE=-libav for ffmpeg, USE=ffmpeg for optional libav/ffmpeg) |
|
24% |
[ 18 ] |
| B (USE=ffmpeg/libav to select, and USE=avcodec for optional either of them) |
|
29% |
[ 22 ] |
| C (FFMPEG_IMPL=ffmpeg/libav to select, USE=ffmpeg for optional either of the impls) |
|
45% |
[ 34 ] |
|
| Total Votes : 74 |
|
| Author |
Message |
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
 Posted: Wed Feb 04, 2015 9:40 am Post subject: |
 |
|
| mv wrote: | | In this thread, I am really ashamed about the community. Gentoo what has become of you? |
Well, since the introduction of systemd *every* decision not to some people's liking comes under suspicion of conspiracy. |
|
| Back to top |
|
 |
tclover
Guru
Joined: 10 Apr 2011
Posts: 516
|
| Posted: Wed Feb 04, 2015 10:26 am Post subject: |
|
|
| genstorm wrote: | | mv wrote: | | In this thread, I am really ashamed about the community. Gentoo what has become of you? |
Well, since the introduction of systemd *every* decision not to some people's liking comes under suspicion of conspiracy. |
Come on, don't make every single Gentoo User out there a _conspi-[geek-]nerd_... I AM NOT ONE for your info!
And do _not_ deny that some dirty tricks were used to force users into... GNOME/SystemD OS $***. Or pilling up unit shit files, for one thing, when using a SystemDebug free OS.
How do you call that? Cheap conspiracy?! Come on. I was not born _after_ SystemD OS,--but before,--nor did I start using a Linux OS or a Gentoo/Linux OS after SystemD OS.
_________________
home/:mkinitramfs-ll/:supervision/:e-gtk-theme/:overlay/ |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6747
|
| Posted: Wed Feb 04, 2015 10:34 am Post subject: |
|
|
| genstorm wrote: | | Well, since the introduction of systemd *every* decision not to some people's liking comes under suspicion of conspiracy. |
But this thread is not about systemd or about a big decision on gentoo's future direction: It is about a small technical detail for a few packages - whether something is split into 2 or 3 USE-flags and of which type - which is really not a big deal and probably not even realized by the majority of users. In fact, it was perhaps even overkill to start even a poll about this. If I were a developer, I would in future make such tiny decisions without RFC if such a poll ends up in a pointless flamewar on rather unrelated topics. |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Wed Feb 04, 2015 10:46 am Post subject: |
|
|
| mv wrote: | | But this thread is not about systemd or about a big decision on gentoo's future direction: It is about a small technical detail for a few packages - whether something is split into 2 or 3 USE-flags and of which type - which is really not a big deal and probably not even realized by the majority of users. |
...and such was the thread about CPU_FLAGS_X86, where we did register the 's-word'. But even without it, the ffmpeg vs. libav topic is already loaded with misinformation and conspiracy theories long enough.
| tclover wrote: | | Come on, don't make every single Gentoo User out there a _conspi-[geek-]nerd_ |
Did I? |
|
| Back to top |
|
|
tclover
Guru
Joined: 10 Apr 2011
Posts: 516
|
| Posted: Wed Feb 04, 2015 10:56 am Post subject: |
|
|
| mv wrote: | One guy not being able to understand why he is making technically impossible "suggestions", despite explanations in the first posting, and even being offended when this is told to him.
Another guy complaining that gentoo allowed choice and attacking the developers that gave him choice and attacking another developer for trying hard to make the choice more convenient and technically sane (please recall: easy choice is among the distribution's main philosophy).
A developer calling this kindergarden what it is and being attacked for doing so.
All in all due to some loud people a decision going overboard which has a clear majority of user's votes.
In this thread, I am really ashamed about the community. Gentoo what has become of you? |
A bunch of morons... What else?!!!
_________________
home/:mkinitramfs-ll/:supervision/:e-gtk-theme/:overlay/ |
|
| Back to top |
|
|
Roman_Gruber
Advocate
Joined: 03 Oct 2006
Posts: 3846
Location: Austro Bavaria
|
| Posted: Wed Feb 04, 2015 11:41 am Post subject: |
|
|
| I do not believe we need such posts at all. Please value the forum rules and be polite |
|
| Back to top |
|
|
khayyam
Watchman
Joined: 07 Jun 2012
Posts: 6227
Location: Room 101
|
| Posted: Wed Feb 04, 2015 2:36 pm Post subject: |
|
|
| mgorny wrote: | | If you cared enough to take a look at the mailing list, you'd know that I already did the work for C, including a new news item and a big diff for all ebuilds. All that is wasted now. Does that make you happier? |
mgorny ... I never take pleasure in someone suffering, so no, there is nothing for me to be happy about, and I don't see how your wasted effort is in any way my fault. When I read "if you cared enough", or "are you happy now", I see nothing but an emotive argument, that's not something I expect in a *reasonable* discussion, its a sign of someone trying to play to the emotions so that their interlocutor is painted as "bad" for having done something that might be attributable as the source of that emotion. In the study of argumentation that's what we call a "black art", and I'm thoroughly immune to it.
| mv wrote: | | One guy not being able to understand why he is making technically impossible "suggestions", despite explanations in the first posting, and even being offended when this is told to him. |
@mv ... fair enough, but he also offered an opportunity to vent on the conspiratorialists.
| mv wrote: | | Another guy complaining that gentoo allowed choice and attacking the developers that gave him choice and attacking another developer for trying hard to make the choice more convenient and technically sane (please recall: easy choice is among the distribution's main philosophy). |
In my defence I'll say that the points I'd raised are pertinent wrt to the fact that effects can be attributable to causes and focusing on the effects in absence of the cause is to miss how we come to arrive at them. If that is an "attack" then that closes off any possibility of discussion, because it is causal relation which provides the subject of the discussion. Also, why am I a complainant? Is anything I've said in that regard non-factual? Please keep in mind the odium theologicum ... disagreement does not equate to an attack, nor does voicing it become a complaint.
| mv wrote: | | A developer calling this kindergarden what it is and being attacked for doing so. |
Sorry, please provide a more exact explication of this narrative ... because I'm not seeing such child like behaviour, or such an attack.
| mv wrote: | | All in all due to some loud people a decision going overboard which has a clear majority of user's votes. |
Yes, but such a poll without the option to register dissension, or deferral, does offer itself to the charge of being a closed ring circus. Its easy to create such a poll ... are you A). a national socialist, or B). a stalinist ... sorry, no third option on offer.
| mv wrote: | | In this thread, I am really ashamed about the community. Gentoo what has become of you? |
I think its necessary we should ask such questions, what is it that will create and sustain us as a community, how do we come to be as we are, are there things that cause its diminution, and if so what, how can they be mitigated against, etc, etc ... these are all *highly* pertinent ... but its only through the lens of causality that we can begin to formulate such questions ... and that is precisely what I was attempting to do. I'm not a naysayer ... I consider the community important, but its for that reason I'm willing to hold it to some standard, and to raise objections where and when I see the reason to do so, that is a contribution, not a complaint or attack.
best ... khay |
|
| Back to top |
|
|
John R. Graham
Administrator
Joined: 08 Mar 2005
Posts: 10587
Location: Somewhere over Atlanta, Georgia
|
| Posted: Wed Feb 04, 2015 3:00 pm Post subject: |
|
|
Folks, please tone it down. Don't you want developers to ask the community's opinion? I certainly do, but I'm afraid that we're discouraging them from doing so. Not locking this thread (yet) as there seems to be some productive discussion still happening.
- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters. |
|
| Back to top |
|
|
sera
Retired Dev
Joined: 29 Feb 2008
Posts: 1017
Location: CET
|
| Posted: Wed Feb 04, 2015 3:18 pm Post subject: |
|
|
| GrueXYZ wrote: | | Am I assuming correctly that this issue is not unlike OpenSSL with several implementations of it (Open, Polar), and one more coming up - LibreSSL? |
It shouldn't be but unfortunately is. In the ssl case the implementations don't block each other and so don't impose a global selection. Consumers can chose which to support without hurting the end-user unreasonably,
| Quote: |
To me the cleanest solution would be for ffmpeg to install/build against ffmpeg, and libav to install/build against libav, and then portage ensuring that both can't be defined. However, that approach would probably make impossible any combinations where one package would build against ffmpeg, and another against libav - if something like that is at all desirable (and imho it is, for choice is good). In which case I really think that a global (or per-package) USE flag of ffmpeg should mean both/any of the two, and then packages define FFMPEG_IMPL, or whatever. |
For implementation reasons it's desirable to keep the number of USE_EXPAND flags as small as possible. Arguably more reasonable suggestion for new USE_EXPAND flags were shot down on this basis. Therefore it's to be expected that this proposal would share that fate. |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Wed Feb 04, 2015 7:08 pm Post subject: |
|
|
| sera wrote: | | It shouldn't be but unfortunately is. In the ssl case the implementations don't block each other and so don't impose a global selection. Consumers can chose which to support without hurting the end-user unreasonably, |
Wouldn't gst-plugins-meta be a good reason to have two separate use flags for this?
FWIW I'm not happy with the current setup for SSL either, where in some places a bare "ssl" gets you OpenSSL and there's a separate "gnutls" flag that also needs setting. There's already a USE_EXPAND for this, but it's specific to curl (and shouldn't be). |
|
| Back to top |
|
|
sera
Retired Dev
Joined: 29 Feb 2008
Posts: 1017
Location: CET
|
| Posted: Fri Feb 06, 2015 12:57 pm Post subject: |
|
|
| Ant P. wrote: | | sera wrote: | | It shouldn't be but unfortunately is. In the ssl case the implementations don't block each other and so don't impose a global selection. Consumers can chose which to support without hurting the end-user unreasonably, |
Wouldn't gst-plugins-meta be a good reason to have two separate use flags for this? |
gst-plugins-meta is indeed mind boggling. Do you mind outlining how you would implement this. I mean just having redundant flags I don't see as an improvement. First I'm not aware of a single such case and secondly you could still end up with ffmpeg if you set USE="- ffmpeg libav" for gst-plugins-meta or vise versa.
| Quote: | | FWIW I'm not happy with the current setup for SSL either, where in some places a bare "ssl" gets you OpenSSL and there's a separate "gnutls" flag that also needs setting. There's already a USE_EXPAND for this, but it's specific to curl (and shouldn't be). |
Lot of people share your sentiment, though in lack of a better solution I'd rather keep it as it is. While ssl is an obvious example of this pattern it's not the only one. opengl vs egl, gtk2 vs gtk3 or X vs wayland are other examples and you certainly can find plenty more. If you can formalize a satisfying scalable solution for this pattern in form of a glep I'm sure it will be considered. Possibly we would want IUSE to be a tree instead of a flat list.
I agree, the curl specific USE_EXPAND flag for ssl is an oddity. |
|
| Back to top |
|
|
petteyg359
n00b
Joined: 30 Apr 2005
Posts: 61
|
| Posted: Fri Feb 06, 2015 11:35 pm Post subject: |
|
|
| sera wrote: | | I agree, the curl specific USE_EXPAND flag for ssl is an oddity. |
A library (e.g. curl) supporting so many different SSL libraries is also an oddity. I'd like more oddity, preferably in the form of GnuTLS and NSS versions of Qt :p
Back to topic, why does there need to be a default? If there's a default, one side or the other is going to have to change their setup. Default to no implementation, and force the user to choose one, until a certain upstream either stops squatting on somebody else's namespace or starts guaranteeing API+ABI compatibility. |
|
| Back to top |
|
|
ct85711
Veteran
Joined: 27 Sep 2005
Posts: 1791
|
| Posted: Sat Feb 07, 2015 1:43 am Post subject: |
|
|
| pettey: the only issue is having no default implementation, is that we end up staying where we are now, in that a lot of packages staying marked, because the two package are diverging to break everything and the devs can't focus on which one to work with. Sadly I doubt this issue is going to be fixed upstream and this is going to continue breaking something at everyone expense. At least with one a default, the devs can make sure everything will at the very least work with one of them, and not keep everything masked. |
|
| Back to top |
|
|
229566
Tux's lil' helper
Joined: 16 Aug 2010
Posts: 127
|
| Posted: Sat Feb 07, 2015 11:59 am Post subject: |
|
|
| sera wrote: | | For implementation reasons it's desirable to keep the number of USE_EXPAND flags as small as possible. Arguably more reasonable suggestion for new USE_EXPAND flags were shot down on this basis. Therefore it's to be expected that this proposal would share that fate. |
Thanks for the explanation. I do wonder then how will LibreSSL be implemented, and that perhaps when deciding about ffmpeg/libav one should think about openssl/libressl as well and have a unified solution? At least with ssl there's room for three-state "ssl openssl libressl" flags, while with ffmpeg there isn't that third one. Eg. with "ssl" you say "enable SSL functionality" having openssl and libressl mutually exclusive and defining which library (interface) to use. Alas, with those two they should be binary compatible so you can do that at the system level. Then again some software is not ready to use LibreSSL yet (eg. Python).
Is it intuitive? Does this result with a mess like having two flags requiring presence of "ssl" and being mutually exclusive? Or is it perhaps redundant as sooner or later Gentoo will switch to LibreSSL and using OpenSSL would be unsupported.
Why am I mentioning ssl in a ffmpeg thread? Because I think it's all part of the same problem: solving for multidimensional package relationships with a flat one-dimensional USE flags list, and that whatever solution is found, it should cover all the known use cases (pun not intended  ).
$.02 |
|
| Back to top |
|
|
mgorny
Developer
Joined: 27 Apr 2007
Posts: 83
|
| Posted: Sat Feb 07, 2015 12:02 pm Post subject: |
|
|
| sera wrote: | | GrueXYZ wrote: | | Am I assuming correctly that this issue is not unlike OpenSSL with several implementations of it (Open, Polar), and one more coming up - LibreSSL? |
It shouldn't be but unfortunately is. In the ssl case the implementations don't block each other and so don't impose a global selection. Consumers can chose which to support without hurting the end-user unreasonably, |
Except for LibreSSL which will collide with OpenSSL and be ABI-incompatible. In fact, the libressl overlay is already switching to a similar setup that we have for ffmpeg/libav.
| Ant P. wrote: | | sera wrote: | | It shouldn't be but unfortunately is. In the ssl case the implementations don't block each other and so don't impose a global selection. Consumers can chose which to support without hurting the end-user unreasonably, |
Wouldn't gst-plugins-meta be a good reason to have two separate use flags for this? |
I don't understand what are you referring to. gst-plugins-libav (pulled in by USE=ffmpeg) supports both ffmpeg & libav.
| sera wrote: | | I agree, the curl specific USE_EXPAND flag for ssl is an oddity. |
It's not an 'oddity', it's a complete design screwup. There're ~4 incompatible curl variants (counting only those the backends that really matter), and some packages require a specific one. I mean, you can already end up being unable to install two packages because they require curl with different SSL backends. |
|
| Back to top |
|
|
petteyg359
n00b
Joined: 30 Apr 2005
Posts: 61
|
| Posted: Sat Feb 07, 2015 5:58 pm Post subject: |
|
|
| mgorny wrote: | | Except for LibreSSL which will collide with OpenSSL and be ABI-incompatible. |
ABI compatibility isn't so bad there, since they're maintaining the API, so the code is still interchangeable (this is Gentoo - just recompile ). libav isn't even trying to keep the same API. |
|
| Back to top |
|
|
mgorny
Developer
Joined: 27 Apr 2007
Posts: 83
|
| Posted: Sat Feb 07, 2015 6:24 pm Post subject: |
|
|
| Oh, I think they don't care about complete API compatibility either. In fact, I myself will try to convince them to kill all those non-sense unsigned char* strings that make C++ development with OpenSSL a PITA. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Gentoo Chat
