[Ffmpeg-devel-irc] ffmpeg-devel.log.20140916

[burek]

[00:03] <iive> so, ffmpeg have been 6 months in the queue, but it shows it have been there for 3 hours debian time. how long have the oldest package been there if it shows 1 year in debian  time?
[00:03] <ubitux> actually probably 1yr because there is only one version for those
[00:05] <ubitux> most packages only have one version so it's fairly safe to assume that ffmpeg is part of the oldest
[00:06] <iive> i'm just kidding... but yeh... 1 year.
[00:06] <ubitux> i count less than 10 packages with multiple versions, and most of them have very close micro numbers
[00:07] <ubitux> so ffmpeg is probably between pos 5 and 10
[00:07] <iive> well... there is nothing showing that... is there?
[00:08] <llogan> anyone planning on making a news entry for 2.4?
[00:11] <llogan> these new headphones aren't as good as i hoped (ATH-M40x). guess it doesn't really matter since they're just for field monitoring.
[00:13] Action: llogan goes back to the Alessandro MS-1
[01:14] <michaelni> llogan, i was planning it but it appears i forgot
[01:23] <michaelni> llogan, just posted a patch for 2.4 news, dont hesitate to comment or correct, ill probably push it soon 
[01:27] <llogan> michaelni: 14th July?
[01:29] <cone-192> ffmpeg.git 03Michael Niedermayer 07master:b76d6132e385: avformat/network: move sockaddr_union after sockaddr_storage compatibility code
[01:30] <michaelni> September ... :)
[01:30] <llogan> otherwise, LGTM.
[01:31] <llogan> thanks. i was thinking of writing one, but I'm at work now. color correcting/grading is so boring.
[01:32] <llogan> ...if you were wondering why i asked but didn't just do it
[04:10] <cone-192> ffmpeg.git 03Michael Niedermayer 07master:aaeae281a898: avformat/mpegtsenc: More verbose warning messages in case of missing video ES startcodes
[04:10] <cone-192> ffmpeg.git 03Michael Niedermayer 07master:688a40b4ed7d: avcodec/h264_sei: ff_h264_decode_sei: dont try to parse trailing zeroes
[07:14] <ubitux> oooh
[07:15] <ubitux> that's a good morning news :o
[07:20] <Timothy__Gu> Good night :)
[07:33] <jamrial> Timothy__Gu: why did you drop the "Discussions about..." part from the channel topic?
[07:33] <Timothy__Gu> jamrial: it's too long. Plus it's already covered in "FFmpeg dev channel"
[07:34] <Timothy__Gu> I really need to sleep now do don't ask me any more questions please. Thanks
[07:36] <Timothy__Gu> https://tracker.debian.org/news/570904 is the key upload BTW
[11:24] <ubitux> should we consider the interfaces with cli (and similar AVOption stuff) as API and bump accordingly the "FFmpeg bundle" when they change?
[11:25] <nevcairiel> isnt the whole avoption stuff specifically designed to avoid issues that API would usually have
[11:25] <ubitux> it's still an API
[11:26] <ubitux> it's not C, but still a user interface that can be used for programming
[11:26] <ubitux> typically ’ scripts
[11:26] <ubitux> and we actually try to keep it by not renaming/delete these options randomly
[11:27] <ubitux> having a 2.4 ’ 3.0 bump could mean that some options somewhere (in whatever libs) got changed
[11:27] <ubitux> of course, we would need to document it
[11:28] <ubitux> (changed = renamed or deleted)
[11:32] <ubitux> it doesn't happen often; the "draw" option in drawtext is one example
[11:38] <cone-798> ffmpeg.git 03Michael Niedermayer 07release/2.4:8ac3b2cdb751: RELEASE_NOTES: add H.261 & HEVC RTP support
[11:39] <ubitux> BTW, unrelated, is there a special tag or whatever marker for "backport relevant"?
[11:44] <michaelni> no but could make sense to add such a thing, what i do is generally backporting stuff locally when i do the work in master 
[11:47] <iive> ubitux: imho, bump only when option is removed or changed/renamed. if old scripts work, no need for bumps.
[11:48] <ubitux> iive: that's what i suggested
[11:49] <iive> but... what number are you going to bump and why?
[11:51] <ubitux> we are currently at 2.4. When we decide to make a new release, we check if we renamed some option or anything user interface relevant (like ffprobe output change or whatever). If so, it will be 3.0, if not 2.5.
[11:52] <ubitux> if we have a FF_API compat layer, we don't need to, until the next bump
[11:58] <iive> so, you propose a special number for cli options?
[11:58] <ubitux> a propose a non-random way of defining the version for the bundle
[11:58] <ubitux> i* propose
[11:59] <ubitux> 1.x ’ 2.x was arbitrary
[11:59] <ubitux> like, hey that's cool let's bump
[11:59] <ubitux> but no meaning whatsoever
[12:00] <ubitux> 2.3 ’ 2.4 no major bump even though we major bump the libraries leading to a API change in drawtext ("draw" option dropped)
[12:09] <nevcairiel> i think any version change in the package is considered "major", as the numbering is arbitrary
[12:10] <cone-798> ffmpeg.git 03Diego Biurrun 07master:2d589273dd36: configure: Split adding of CFLAGS and CPPFLAGS for hardened toolchain
[12:10] <cone-798> ffmpeg.git 03Michael Niedermayer 07master:5ddfac81c4a9: Merge commit '2d589273dd36c5eb271a035ea0e669b64dae257f'
[13:01] <wm4> ubitux: I'd consider AVOptions API
[13:01] <wm4> definitely
[13:02] <ubitux> we already do, in the library themselves
[13:02] <ubitux> but for example, a major bump might *not* cause an api user bump from the CLI PoV
[14:09] <wm4> so ffmpeg 2.4 made it into debian experimental
[14:09] <iive> yes it finally did.
[14:25] <ubitux> libav distributing ffmpeg tshirts? we need to print libav tshirts to trade with them!
[14:26] <wm4> they're violating your trademark rights, you must start thermonuclear war with them
[14:32] <iive> well send polite green men first. they are very polite.
[14:59] <cone-798> ffmpeg.git 03Stefano Sabatini 07master:3a0c70f1e374: doc/examples/transcoding: use av_packet_rescale_ts()
[16:08] <thardin> does anyone in here know what kind of convention DFD is?
[16:17] <J_Darnley> No.  Any more context?
[16:18] <av500> http://en.wikipedia.org/wiki/Dog_Fashion_Disco
[16:19] <J_Darnley> "Donald Fauntleroy Duck, the full name of Donald Duck"
[16:19] <nevcairiel> he has a middle name o.o
[16:19] <nevcairiel> fascinating
[16:26] <thardin> J_Darnley: not really. a friend of mine heard of the acronym
[16:27] <nevcairiel> in multimedia i only know DSD, DXD and DVD
[16:32] <cone-798> ffmpeg.git 03Steven Liu 07master:2c21e8b1a490: lavf/hls: add hls_ts_option
[16:57] <cone-798> ffmpeg.git 03Clément BSsch 07master:de0e49c2450f: avformat/flacenc: use av_clip() instead of av_clip_c()
[16:57] <cone-798> ffmpeg.git 03Clément BSsch 07master:d469aa8cfaa9: sws: use av_clip() instead of av_clip_c()
[18:26] <cone-798> ffmpeg.git 03Michael Niedermayer 07master:4a99134f1a71: tools/crypto_bench: fix build when AV_READ_TIME is unavailable
[18:59] <cone-798> ffmpeg.git 03Carl Eugen Hoyos 07master:7c3ba1b83aeb: Add a closing LF to Webm-manifest files.
[18:59] <cone-798> ffmpeg.git 03Michael Niedermayer 07master:a07b0ffdc342: Merge remote-tracking branch 'cehoyos/master'
[19:07] <abaudoux> Hello, I have the following error when demuxing a ts file : http://pastebin.com/JmUXNhWz
[19:55] <compn> Abaudoux
[19:55] <compn> Oh he gone
[20:19] <cone-798> ffmpeg.git 03Mikulas Patocka 07master:9030c58a780a: configure: Disable i686 for i586 and lower CPUs
[20:19] <cone-798> ffmpeg.git 03Michael Niedermayer 07master:9bf260b2d2fc: Merge commit '9030c58a780a02fb8256cb75cd39275eaad786d1'
[20:38] <cone-798> ffmpeg.git 03Gabriel Dume 07master:45ff7c93dd84: dca: K&R formatting cosmetics
[20:38] <cone-798> ffmpeg.git 03Michael Niedermayer 07master:7d43fbe3aef8: Merge commit '45ff7c93dd84a254cc96acc589e5ac3d7bd16bce'
[21:24] <cone-798> ffmpeg.git 03Clément BSsch 07master:a87527ad6737: avfilter/ass: make sure the log level are in available range
[21:24] <cone-798> ffmpeg.git 03Clément BSsch 07master:8a9c5db29e84: avfilter/ass: better log level mapping
[21:24] <cone-798> ffmpeg.git 03Clément BSsch 07master:4b58349bc8ff: avfilter/ass: add shaping option
[21:25] <ubitux> JEEB: saw that https://trac.ffmpeg.org/ticket/3949#comment:2 ?
[21:26] <JEEB> nope, but I will check it when I'm not tired as fuck
[21:30] <ubitux> ok :)
[21:36] <ubitux> llogan: i don't understand the second example
[21:37] <llogan> ubitux: would an image displaying what (I expect) it does help?
[21:43] <llogan> ffmpeg -f lavfi -i testsrc=s=hd720:d=5 -filter_complex "geq=lum=255*(Y/H),format=gray[grad];[0:v]boxblur=4[blur];[blur][grad]alphamerge[alpha];[0:v][alpha]overlay" output.mp4
[21:44] <llogan> it just gradually gets blurry. not blurry on top, blurry on bottom.
[21:44] <ubitux> say that in the description please
[21:46] <llogan> words hard.
[21:47] <llogan> i'll come up with something
[21:47] <llogan> how about? "In this example the video will progressively become more blurry along the y-axis as defined by the linear gradient."
[21:48] <llogan> or "...more blurry from top to bottom."
[21:48] <llogan> if it's a shitty example i can remove it
[21:51] <ubitux> alright, so this says what it will look like, now add how it's done
[21:51] <ubitux> and i'll be fine with that ;)
[21:51] <llogan> alright. thanks for taking a look at it.
[21:54] <llogan> wm4: elementary appears to be the least ugly of the ugly claws themes
[21:54] <wm4> claws as themes?
[21:54] <wm4> *has
[21:54] <llogan> http://www.claws-mail.org/themes.php
[21:55] <wm4> ...
[23:43] <cone-798> ffmpeg.git 03Michael Niedermayer 07master:474d858fd955: Revert "avformat/rtpproto: fix strict aliasing violations with sockaddr"
[00:00] --- Wed Sep 17 2014