[Ffmpeg-devel-irc] ffmpeg-devel.log.20190513

[burek]

[01:40:49 CEST] <taliho> ##fastCoder
[01:41:21 CEST] <taliho> sorry didn't mean to post that
[07:05:43 CEST] <cone-783> ffmpeg 03Gyan Doshi 07master:87db1ca632bf: avfilter/drawtext: make command processing error-resilient
[08:53:51 CEST] <cone-783> ffmpeg 03Gyan Doshi 07master:dcc999819dda: doc/filters: mention drawtext command support
[12:16:29 CEST] <cone-471> ffmpeg 03Paul B Mahol 07master:f1c9d6fe704c: avfilter/vf_chromahold: implement blend option
[12:19:44 CEST] <durandal_1707> how to boost commits counts?
[12:24:15 CEST] <kierank> durandal_1707: work on some hardware decoding
[12:24:24 CEST] <kierank> one commit per codec ^10 hardware decoders
[12:59:40 CEST] <durandal_1707> what about this ndi voting, what are results ? where is transparency?
[13:08:08 CEST] <durandal_1707> hello? is anybody here?
[13:17:27 CEST] <durandal_1707> nobody. no transparency. no objective thinking. dark times.
[13:18:55 CEST] <TheAMM> I don't think it's fair to expect everyone be around at all times?
[13:20:56 CEST] <durandal_1707> nonsense, they are connected all the time
[13:35:57 CEST] <cehoyos> durandal_1707: Which filter do you mean?
[13:36:01 CEST] <cehoyos> for example
[13:36:33 CEST] <durandal_1707> cehoyos: whenever i write filter, i copy it from michaelni filter
[13:36:42 CEST] <cehoyos> I don't think so.
[13:54:44 CEST] <BBB> durandal_1707: I think ndi will be summarized by thilo when he's ready
[13:54:56 CEST] <BBB> feel free to poke him about it
[13:55:31 CEST] <cehoyos> In what way was Thilo involved?
[14:10:12 CEST] <cone-471> ffmpeg 03Adam Richter 07master:b8ed4930618b: libswcale: Fix possible string overflow in test.
[14:10:13 CEST] <cone-471> ffmpeg 03Michael Niedermayer 07master:9d269301f017: swscale/tests/swscale: Lengthen pixfmt name buffer to 21 bytes
[15:21:47 CEST] <BBB> cehoyos: sorry, my bad, it was marton who organized the vote, so s/thilo/marton/ in that sentence
[15:22:23 CEST] <BBB> and it is indeed approximately 14 days since the vote, so I would probably expect it by today or sometime this week
[15:29:47 CEST] <cehoyos> Ther are one or two things to mention:
[15:30:24 CEST] <cehoyos> After posing one question, that (imo really) still makes no sense no matter how often I read it, and a second question that everybody agreed was not such a good idea, he gave us seven days to answer.
[15:30:31 CEST] <cehoyos> After ten days, he answered himself.
[15:30:54 CEST] <cehoyos> All-in-all, I am not sure if this vote will be remembered as a particular well-done solution.
[15:32:32 CEST] <cehoyos> s/answer/vote
[15:37:54 CEST] <BBB> ¯\_(Ä)_/¯
[17:04:30 CEST] <mkver> @philipl: Is it possible that you forgot to commit the reference files for the filter-pixdesc-nv24/42 fate tests?
[18:30:55 CEST] <cone-471> ffmpeg 03Gyan Doshi 07master:21832b93d533: doc/filters: add scale2ref example for proportional scaling
[20:52:25 CEST] <cone-471> ffmpeg 03Andreas Rheinhardt 07release/3.2:9fad760f5648: lavf/webm_chunk: Respect buffer size
[20:52:26 CEST] <cone-471> ffmpeg 03Michael Niedermayer 07release/3.2:a649b62b903c: avcodec/hq_hqa: Check available space before reading slice offsets
[20:52:27 CEST] <cone-471> ffmpeg 03Adam Richter 07release/3.2:280f5c4fcfc9: libswcale: Fix possible string overflow in test.
[20:52:28 CEST] <cone-471> ffmpeg 03Michael Niedermayer 07release/3.2:3a6bcc059cc6: Changelog: update
[22:17:20 CEST] <BBB> cehoyos: I think you're being passive agressive at this point. if you feel that the vote also means we remove ndi from release branches, why not submit a patch?
[22:17:44 CEST] <BBB> your response feels like an attack on the voting process and/or marton, which is not necessary at all
[22:17:51 CEST] <cehoyos> As said before, I still wonder how to interpret "should libndi be removed" differently...
[22:18:17 CEST] <cehoyos> How should I not attack him? He said the vote would take seven days, but he voted after ten days.
[22:18:38 CEST] <cehoyos> (I know you have forgotten a rather important vote in this project, I - sadly - haven't)
[22:18:40 CEST] <BBB> he later correct and said it'd be 14 days
[22:18:47 CEST] <cehoyos> I missed that.
[22:18:47 CEST] <BBB> stop attacking me
[22:18:57 CEST] <BBB> be open, be constructive
[22:19:04 CEST] <BBB> don't think everyone except you is an ass
[22:19:06 CEST] <BBB> I'm not an ass
[22:19:09 CEST] <BBB> we all have the best intentions
[22:19:13 CEST] <cehoyos> Writing that I am passive aggressive seems not like a compliment, no?
[22:19:13 CEST] <BBB> try to start with that
[22:19:17 CEST] <BBB> and everything will be better
[22:19:22 CEST] <cehoyos> Yes, please do;-)
[22:19:28 CEST] <BBB> <3
[22:20:14 CEST] <cehoyos> Again: How could "should libndi be removed" be interpreted? I am open for suggestions, I just learned this week that "not reacting at all" was the preferred option for one developer here.
[22:20:25 CEST] <BBB> and that's fine
[22:20:26 CEST] <cehoyos> (the two are of course not related)
[22:20:52 CEST] <cehoyos> No, sorry, the question was not "is it fine or not" but "how could this be interpreted?"
[22:21:02 CEST] <BBB> it is fine that some of us do not want to vote
[22:21:30 CEST] <BBB> I sometimes want to be able to not react at all. it's blissful
[22:21:58 CEST] <BBB> how to interpret it? in whatever way you want, as long as it conforms to the sentence "libndi support is removed"
[22:22:11 CEST] <BBB> if you feel that "in release branch" conforms to that statement, then send a patch
[22:22:33 CEST] <BBB> it sounds like an awesome contribution. I totally would support such a patch, even if it breaks a """feature""" in a release branch
[22:22:49 CEST] <BBB> """feature""" intentionally in quotes, because I consider ndi a bug
[22:23:40 CEST] <cehoyos> Of course! I didn't vote myself.
[22:24:11 CEST] <cehoyos> I find the claim that we would break distribution when removing a non-free library funny btw...
[22:24:17 CEST] <cehoyos> distributions
[22:24:40 CEST] <cehoyos> I still wonder what the question actually meant
[22:39:57 CEST] <nevcairiel> I thought release branches don't even exist in your mind anyway :)
[22:41:14 CEST] <cehoyos> Catch!
[22:49:46 CEST] <beastd> I'm not sure about the fine language lawyer details, but IMHO Marton made his intent clear when he clarified that he wants the vote to confirm or not confirm the commit that removed libndi from master
[22:50:10 CEST] <beastd> This email http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2019-April/243324.html
[22:52:05 CEST] <BradleyS> lots of great engineers here, maybe this is a sign it's time to develop a voting platform ;)
[22:52:54 CEST] <kierank> 21:24:13 <"cehoyos> I find the claim that we would break distribution when removing a non-free library funny btw...
[22:53:06 CEST] <kierank> I agree but if people build releases themselves we shouldn't break that
[22:53:41 CEST] <cehoyos> We accept --enable-avconv
[22:53:48 CEST] <cehoyos> (At least we did for a very long time)
[22:53:55 CEST] <cehoyos> We could silently accepts --enable-libndi
[22:54:13 CEST] <BradleyS> from what i gather, the issue is less about whether ndi should be removed, and more about whether it should be in --enable-nonfree and *not* built by default into release builds and distro builds
[22:54:34 CEST] <cehoyos> beastd: I did not find it clear (but I arguably found a few in the >50 mails that I already had forgotten about)
[22:54:46 CEST] <BradleyS> the latter would allow affected parties to build themselves as long as they don't redistribute
[22:54:47 CEST] <cehoyos> BradleyS: No
[22:54:58 CEST] <cehoyos> It was already removed and was always nonfree
[22:55:05 CEST] <cehoyos> (before it was removed)
[22:55:08 CEST] <BradleyS> some would probably find that not good enough given the license infringement
[22:55:18 CEST] <BradleyS> yes, but it was built by default and released, no?
[22:55:19 CEST] <cehoyos> That it was removed?
[22:55:38 CEST] <cehoyos> No, nonfree libraries cannot be build by default, it would make our binaries useless...
[22:55:41 CEST] <BradleyS> *some would probably find keeping it in source but not binary not enough
[22:55:50 CEST] <BradleyS> oh right, i'm tired.
[22:55:51 CEST] <cehoyos> I mean, not useless, but distros would have issues with them
[22:56:17 CEST] <BradleyS> ndi should really open the source and api, problem solved
[22:56:23 CEST] <cehoyos> Of course.
[22:56:23 CEST] <BradleyS> it's not like they're going to lose hardware business
[22:56:31 CEST] <BradleyS> they must have awful lawyers
[22:56:50 CEST] <cehoyos> This is the alternative that I hoped to hear when asking for alternatives, but nobody suggested it (before you did)
[22:57:06 CEST] <cehoyos> I am not 100% sure they have lawyers...
[22:57:19 CEST] <BradleyS> it's a great alternative but not something ffmpeg can make happen directly of course
[22:57:38 CEST] <cehoyos> Probably, since they threatened somebody, but if they had an active lawyer all this likely would not have happened
[22:57:49 CEST] <cehoyos> Suing is not an alternative?
[22:58:14 CEST] <BradleyS> i think j-b intends to
[22:58:23 CEST] <BradleyS> or at least he's indicated
[22:58:38 CEST] <cehoyos> If he had said that, we would not have had to remove the library imo...
[22:58:43 CEST] <kierank> BradleyS: they will lose vendor lockin if they do that
[22:58:44 CEST] <cehoyos> If he had told us
[22:59:01 CEST] <BradleyS> kierank: possibly in software but not hardware necessarily
[22:59:15 CEST] <kierank> BradleyS: it allows people to build their own hardware if the source is open
[22:59:20 CEST] <BradleyS> unless someone else wants to build on it significantly; it's a serious investment
[22:59:21 CEST] <BradleyS> yeah
[22:59:23 CEST] <kierank> without a licence
[22:59:32 CEST] <BradleyS> not like it's difficult to re anyway
[22:59:40 CEST] <kierank> then they send legal threats
[22:59:51 CEST] <BradleyS> building an open source alternative that's feature complete would be great
[22:59:57 CEST] <cehoyos> With the lawyers who do not read their blogs;-)
[22:59:58 CEST] <kierank> and nobody will use it
[23:00:13 CEST] <kierank> cehoyos: no carl, they just don't care about ffmpeg copyright, only theirs
[23:00:33 CEST] <beastd> cehoyos: i agree the initial mail of the vote is not worded accurately and that's a bit problematic because in the end one votes on the initial question.
[23:00:35 CEST] <BradleyS> whatever happens, it seems voting on a mailing list in the way that's occurred is less than ideal
[23:00:36 CEST] <BBB> BradleyS: I don't agree btw, I believe --enable-nonfree was for gpl-incompatible stuff like the free-but-not-gpl-compatible aac encoder etc.
[23:00:38 CEST] <cehoyos> This is of course a very possible alternative
[23:00:47 CEST] <BBB> BradleyS: --enable-nonfree being used for closed-source stuff is a travesty
[23:00:47 CEST] <cehoyos> beastd: Thank you!
[23:01:04 CEST] <BBB> we are not a closed-source project; we are an opensource project; we encourage people to use opensource solutions
[23:01:10 CEST] <cehoyos> BBB: It was used for nero, please...
[23:01:11 CEST] <BradleyS> fair enough, should rename to --enable-nongpl
[23:01:15 CEST] <durandal_1707> please add --enable-proprietary
[23:01:17 CEST] <JEEB> yea, one can take a look at the option's history
[23:01:25 CEST] <BBB> BradleyS: maybe
[23:01:26 CEST] <JEEB> BradleyS: I really agree the naming is not the best indeed :)
[23:01:35 CEST] <BBB> durandal_1707: no
[23:01:48 CEST] <cehoyos> kieran is right, the name is 100% clear for those who want to understand
[23:01:49 CEST] <BBB> durandal_1707: unless whatever company needs it pays me ONE MILLION DOLLARS
[23:01:50 CEST] <BradleyS> maybe that's the best next step, split --enable-nonfree into --enable-nongpl and --enable-proprietary
[23:02:01 CEST] <BBB> was that the fifth element?
[23:02:01 CEST] <BradleyS> then vote on removing --enable-proprietary :P
[23:02:12 CEST] <cehoyos> nero?
[23:02:19 CEST] <cehoyos> Of libfaac?
[23:02:33 CEST] <cehoyos> Or
[23:02:37 CEST] <JEEB> it was first added for libamr-nb
[23:02:40 CEST] <JEEB> IIRC
[23:02:51 CEST] <JEEB> then faac got caught using incompatible reference implementation code
[23:02:53 CEST] <kierank> cehoyos: nero?
[23:02:53 CEST] <cehoyos> Then nero may have been the fifth;-)
[23:02:59 CEST] <cehoyos> aac encoder
[23:03:00 CEST] <kierank> afaik nero never had a public api
[23:03:03 CEST] <cehoyos> proprietary
[23:03:04 CEST] <kierank> it was just some exe they shipped
[23:03:08 CEST] <cehoyos> We used it
[23:03:13 CEST] <JEEB> yea they didn't. the exe got used a lot
[23:03:15 CEST] <JEEB> back in ye olden days
[23:03:37 CEST] <kierank> I don't remember nero code being used like that
[23:03:37 CEST] <cehoyos> So it wasn't nero but another encoder?
[23:03:43 CEST] <kierank> because afaik there was never a nero api
[23:03:54 CEST] <kierank> it was faad and faac
[23:04:01 CEST] <cehoyos> I may misremember but there was another closed source encoder imo
[23:04:02 CEST] <kierank> which had public code but not gpl compat
[23:04:07 CEST] <kierank> I don't recall this
[23:04:09 CEST] <cehoyos> faad is compatible
[23:04:15 CEST] <BBB> right, as I've always said, it was opensource but not gpl-compatible
[23:04:23 CEST] <JEEB> yup
[23:04:42 CEST] <kierank> cehoyos: from my recollection ffmpeg never had proprietary stuff, people used mplayer for the dlls
[23:04:42 CEST] <JEEB> the AMR reference implementation you couldn't distribute or so, which is why it became the first if I recall correctly
[23:05:08 CEST] <JEEB> (but it was open source as in source available)
[23:06:21 CEST] <BradleyS> mit/bsd all the things
[23:06:24 CEST] Action: BradleyS runs
[23:06:27 CEST] <beastd> BradleyS: I don't think that voting on the mailing list was the problem. Though I have no solutions for the problems. Sometimes you come up with things that you think are fit for voting and discover after the fact they aren't.
[23:06:41 CEST] <BradleyS> this is true
[23:06:56 CEST] <BradleyS> i guess my thinking is an actual voting platform would better guide the process
[23:07:04 CEST] <BradleyS> and prevent some of the confusion, etc.
[23:07:59 CEST] <BradleyS> for instance, fields to clearly define the topic, what happens specificially with a yea/nay vote, a period for comments/discussion, and a clear voting period with votes being strictly yea/nay on each count
[23:08:06 CEST] <BradleyS> more rigidity
[23:08:59 CEST] <beastd> Maybe we should leave some time before a vote actually is effective. people can use that time to ask questions about the questions/points of the vote
[23:09:11 CEST] <kierank> BradleyS: it really won't, this is the same technical solution to a social problem that we discuss every year
[23:09:37 CEST] <thardin> votes suck tbh
[23:09:52 CEST] <JEEB> what it showed was basically that people have differing opinions on what the unclearly named option meant to begin with. and then of course people could have differing opinions on the original action that spurred the vote (which can be separate from the opinion on the end result).
[23:10:15 CEST] <BradleyS> an explicit period for comments followed by comments being closed, and during the voting period the system only accepts yea/nay, you don't think this would help guide the process?
[23:11:03 CEST] <cehoyos> Of course, but the remaining question is: What would we have done if a majority had voted against "removing" something that was removed earlier?
[23:11:17 CEST] <nevcairiel> Restored it, obviously.
[23:11:21 CEST] <cehoyos> kierank: I misremembered
[23:11:29 CEST] <cehoyos> But that wasn
[23:11:37 CEST] <BradleyS> that would be explicitly defined in my fantasy voting platform :) but probably revert the removal
[23:11:38 CEST] <cehoyos> But that wasn't the question, was it?
[23:12:01 CEST] <nevcairiel> You are just trying to find weird reasons for everthing to be bad and wrong
[23:12:07 CEST] <nevcairiel> this was obvious to everyone else
[23:12:32 CEST] <cehoyos> I would love libndi not to have distributed binaries and still be in the sourc code;-(
[23:12:45 CEST] <BradleyS> cehoyos: btw, j-b did explicitly state "i will sue" on this channel, perhaps before you rejoined some days ago
[23:13:24 CEST] <cehoyos> But adding it back after removing it for a copyright violation is outside of my imagination.
[23:13:33 CEST] <cehoyos> I usually read it but may have missed it, thank you
[23:14:40 CEST] <BradleyS> actually you were here
[23:14:46 CEST] <BradleyS> may 3
[23:14:48 CEST] <BradleyS> [14:10:25] <j-b> They are also violating the copyright of the VideoLAN project.
[23:14:49 CEST] <BradleyS> [14:10:27] <cehoyos> As said, I hope this channel is still free-speech!
[23:14:49 CEST] <BradleyS> [14:10:30] <j-b> and I'm not happy
[23:14:49 CEST] <BradleyS> [14:10:45] <j-b> and I will sue.
[23:14:58 CEST] <BradleyS> my times are UTC-4
[23:15:42 CEST] <durandal_1707> nothing bad happened!
[23:17:08 CEST] <durandal_1707> why you all insist on sueing?
[23:17:17 CEST] <BradleyS> only one person is insisting
[23:18:40 CEST] <cehoyos> It appears we should have waited longer;-)
[23:21:31 CEST] <thardin> some of the stuff NDI does seems p neat. hopefully someone manages to RE it
[23:24:24 CEST] <thardin> I think it's possible to do the same thing its codec does but with h.264, and probably vp* too
[23:25:25 CEST] <JEEB> I think it was mostly control messages/finding devices on network/pulling&pushing from/to devices and then some A/V format that gets streamed
[23:27:08 CEST] <thardin> that too
[23:42:08 CEST] <j-b> BradleyS: how is that a problem?
[23:44:40 CEST] <BradleyS> not suggesting it is, seems legit to me
[23:45:29 CEST] <BradleyS> paul said "you all", was just saying it's probably not everyone's opinion
[23:45:59 CEST] <j-b> There are several issues and people like to confuse them to help muddy the water
[23:46:18 CEST] <pross> re vote: does the company at question even care about the outcome? 
[23:46:23 CEST] <BradleyS> yeah, i think most of the convo is in good faith but it can get muddy fast
[23:46:36 CEST] <j-b> - the redistribution of a non-free FFmpeg version by a company
[23:46:58 CEST] <j-b> - this same company suing several FLOSS developers when they tried to RE the library
[23:47:07 CEST] <j-b> - the violation of VideoLAN copyright
[23:47:15 CEST] <j-b> Those 3 are separate issues.
[23:47:23 CEST] <BradleyS> absolutely
[23:47:38 CEST] <BradleyS> too bad there's no law against being a dick, that would make it 4
[23:47:45 CEST] <j-b> - On the first, it was voted, so it's over
[23:48:05 CEST] <j-b> - On the second, we all know that Newtek are scumbags now;
[23:48:16 CEST] <j-b> - On the third, this is ongoing, and unrelated to FFmpeg
[23:48:22 CEST] <BradleyS> are they actively suing individuals right now? or just threatening
[23:48:55 CEST] <j-b> When you send a cease-and-desist letter, is that suing or threatening ?
[23:49:02 CEST] <j-b> When does "suing" start?
[23:49:09 CEST] <j-b> depends heavily on the juridiction, IIRC
[23:49:21 CEST] <thardin> it's threatening to sue at least
[23:49:35 CEST] <BradleyS> in the us suing is when you file legal papers, but a cnd is a strong indication that could happen
[23:50:28 CEST] <BradleyS> the cnd and response would certainly be evidence in any proceeding
[23:50:35 CEST] <BradleyS> ianal
[23:51:01 CEST] <j-b> So it's a bit more than just threaten.
[23:51:01 CEST] <thardin> you said anal
[23:51:16 CEST] <BradleyS> in any case, the ndi people are wrong across the board, that seems clear enough
[23:51:25 CEST] <BradleyS> i did!
[23:51:47 CEST] <j-b> They don't care: they sold to Vizrt a few weeks ago, and made millioins
[23:52:17 CEST] <thardin> vizrt eh? I know a bunch of people there
[23:53:29 CEST] <BradleyS> i think the consensus on this side of the pond is that a cnd is a legal threat with intent to follow through (as opposed to a verbal threat/warning or email not from legal)
[00:00:00 CEST] --- Tue May 14 2019