[Ffmpeg-devel-irc] ffmpeg-devel.log.20150807

burek burek021 at gmail.com
Sat Aug 8 02:05:02 CEST 2015

[00:00:25 CEST] <debianuser> Yeah, but it looks like they haven't thought it would cause so much noise. They're just kind of friends with Pocket developer: http://www.planet-libre.org/?post_id=18514 - "we've know Nate for a very long time". And they wanted to deliver a feature they thought to be useful...
[01:06:47 CEST] <haasn> debianuser: most of the concerns would be placated by making the server free and open source, and making it so you can self-host your own version
[01:06:50 CEST] <haasn> (same for sync)
[03:31:59 CEST] <cone-384> ffmpeg 03Michael Niedermayer 07master:bee2e702ccd8: tests/fate/mp3: increase mp3-float-extra_overread FUZZ for ppc64be-RHEL7.0-gcc-4.8.2-ibmcrl to 23 from 20
[04:43:49 CEST] <haasn> another silly thing about the whole WebP situation is the fact that WebM exploded onto the internet when it came out, and how every browser has support for it. There's literally no reason WebP couldn't do the same, if it provides tangible benefits over JPEG (which it does)
[05:10:08 CEST] <cone-384> ffmpeg 03James Almer 07master:9c0407e8563f: x86/sbrdsp: remove an unnecessary mova in sbr_autocorrelate
[06:05:08 CEST] <doublebackslash> hey, got a quick question on. Is the use of alloca acceptable or would it not compile / break in some environments?
[06:06:26 CEST] <doublebackslash> I'm just using it for some small pointer arrays in an audio filter. The thinking is that there is no need to invoke malloc for such small amounts of memory but I'm not sure if that is acceptable in mainline
[06:13:30 CEST] <doublebackslash> not sure if quiet this time of night or barking up the wrong tree...
[06:16:30 CEST] <c_14> doublebackslash: probably just quiet. wait around or ask on the ml
[06:17:52 CEST] <doublebackslash> fair enough. I'll switch to av_malloc for now or persist the alloctions between invocations via the priv struct
[06:18:08 CEST] <doublebackslash> can't be worse than breakign a build :p
[06:27:57 CEST] <doublebackslash> okay, this one is a bit more ffmpeg related than memory. Taking a cue from amix I'm using samples as a PTS. It works for me but I'm worried I might be missing a trick. Any fancier / more correct way to handle this?
[06:31:49 CEST] <rcombs> I think alloca is usually frowned upon
[06:32:10 CEST] <rcombs> but I'm not the arbiter of ffmpeg memory allocation calls
[06:32:53 CEST] <doublebackslash> when in doubt do as the Romans do
[06:36:44 CEST] <rcombs> (the romans appear not to use alloca)
[07:42:41 CEST] <durandal_1707> doublebackslash: what filter does?
[09:25:56 CEST] <cone-790> ffmpeg 03Rostislav Pehlivanov 07master:c47c781e83fd: aacenc: Move local encoder specific tables to a separate file
[09:28:18 CEST] <cone-790> ffmpeg 03Alexandra Hájková 07master:944f60866f50: asfdec: read values properly
[09:28:19 CEST] <cone-790> ffmpeg 03Luca Barbato 07master:e5997152f54f: asf: Use time_t where needed
[09:28:20 CEST] <cone-790> ffmpeg 03Hendrik Leppkes 07master:b54cf4b8863d: Merge commit '944f60866f507e3c0850ae9c2f30dac1da54587c'
[09:28:21 CEST] <cone-790> ffmpeg 03Hendrik Leppkes 07master:3d2d6728288f: Merge commit 'e5997152f54f790229c99f237f8eb6b5b1ee683a'
[09:56:03 CEST] <cone-790> ffmpeg 03Jake Sebastian-Jones 07master:9469370fb326: h264: Use AVERROR return codes instead of -1
[09:56:04 CEST] <cone-790> ffmpeg 03Hendrik Leppkes 07master:8015150f4387: Merge commit '9469370fb32679352e66826daf77bdd2e6f067b5'
[16:02:44 CEST] <ubitux> hum, avio_close_dyn_buf doesn't allow a call with a NULL pbuf so that we don't have to get the pointer and free it afterward?
[16:40:50 CEST] <BBB> does anyone have chat logs from libav-devel from the past few years?
[16:41:22 CEST] <nevcairiel> technically yes, i guess, but i would have to build a script to extract them from the database they are stored in
[16:41:47 CEST] <nevcairiel> my irc client doesnt forget
[16:41:48 CEST] <nevcairiel> :d
[16:42:57 CEST] <ubitux> BBB: i have 2012 to beginning 2015 in irssi format
[16:43:08 CEST] <ubitux> rest of 2015 in weechat format
[16:43:46 CEST] <ubitux> with some random holes because of network troubles & stuff i guess
[16:45:30 CEST] <kierank> I have logs too for a similar perioid
[16:46:01 CEST] <BBB> does anyone have them from 2011?
[16:46:05 CEST] <Gramner> since 2011-03-22
[16:46:39 CEST] <BBB> thats probably as early as exists
[16:46:43 CEST] <BBB> when did the channel start?
[16:48:09 CEST] <cone-790> ffmpeg 03Paul B Mahol 07master:2bdd026b264e: avfilter/avf_avectorscope: stop making output fully transparent
[16:48:23 CEST] <Gramner> Registered : Mar 13 16:16:37 2011
[16:48:55 CEST] <BBB> cool, thats about as good as itll get then
[17:12:18 CEST] <Daemon404> i was gonna ask why nevcairiel hasnt merged libav recently... but then i saw libav has had one commit in liek 2 months
[17:12:56 CEST] <Daemon404> ... nope firefox just autocomplete teh wronh url
[17:13:10 CEST] Action: Daemon404 asks nevcairiel about merges
[17:13:15 CEST] <BBB> lol
[17:15:27 CEST] <ubitux> Daemon404: well, only one commit, no?
[17:15:59 CEST] <ubitux> which was actually fixed in the merge commit itself
[17:16:04 CEST] <Daemon404> hmm right
[17:16:04 CEST] <Daemon404> ok
[17:16:17 CEST] <Daemon404> [16:15] <@ubitux> which was actually fixed in the merge commit itself <-- nooooooooo dont start this again
[17:16:39 CEST] <ubitux> well the commit was wrong so it had to be fixed
[17:16:43 CEST] <ubitux> the merge commit is here for that
[17:16:54 CEST] <ubitux> it's more like partially merged
[17:16:56 CEST] <ubitux> instead of "fixed"
[17:16:59 CEST] <Daemon404> its absolutely infuriating to try and bisect and find the change *in* a merge commit
[17:17:08 CEST] <ubitux> git show 8015150f43875531f05e31ac5021ea676fb20e40
[17:17:36 CEST] <ubitux> git log ..libav/master --oneline to see how much difference with libav
[17:17:56 CEST] <Daemon404> heh
[17:25:43 CEST] <nevcairiel> Daemon404: wah, i did merge this morning
[17:25:53 CEST] <Daemon404> i was mistaken
[17:26:03 CEST] <Gramner> would using cherry-pick/rebase instead of merge be an option?
[17:26:10 CEST] <Gramner> if you don't like merge commits
[17:26:25 CEST] <nevcairiel> nah that would be annoying
[17:26:36 CEST] <nevcairiel> merging has the advantage that git tracks for you which changes are not in yet
[17:27:03 CEST] <Daemon404> yeah cheryr pick is a bad road
[17:29:26 CEST] <BBB> ok I got irc logs for devel channels
[17:29:34 CEST] <BBB> does anyone here reside in #libav and #ffmpeg?
[17:29:41 CEST] <BBB> (with logs going back to 2010-2011 or so)
[17:30:43 CEST] <Daemon404> i only have the devel channels
[17:30:46 CEST] <Daemon404> back to 2010/2011
[17:30:53 CEST] <BBB> I already have those
[17:30:59 CEST] <Daemon404> i know
[17:31:16 CEST] <BBB> maybe I should sit in #libav/#ffmpeg and ask there
[17:45:52 CEST] <c_14> BBB: this should have most of the #ffmpeg logs http://ffmpeg.gusari.org/irclogs/
[17:46:20 CEST] <BBB> I think I have that one, its from nov 2011 onwards
[17:46:27 CEST] <BBB> Im hoping to get 2010-2011 also, if possible
[17:46:42 CEST] <Daemon404> 2010 wouldnt exist for libav
[17:47:12 CEST] <BBB> right, thatd (hopefully) be from 2011 (march?) onwards
[17:47:55 CEST] <Daemon404> $ head -1 \#libav-devel.log
[17:47:56 CEST] <Daemon404> --- Log opened Thu Apr 07 10:41:52 2011
[17:48:06 CEST] <Daemon404> for me
[17:48:36 CEST] <BBB> gramner had them starting in march, which seems to be when the channel got registered
[17:48:44 CEST] <Daemon404> ah
[17:48:46 CEST] <BBB> so I suppose thats the earliest available logs
[17:48:54 CEST] <Daemon404> why do you want logs anyway?
[17:54:56 CEST] <JEEB> I might have #ffmpeg logs, since I was masochistic enough to stay around
[17:55:03 CEST] <JEEB> #libav I often forgot to join
[17:56:02 CEST] <BBB> JEEB: would you be willing to share logs with me? or statistics on logs
[17:56:12 CEST] <BBB> youre on 24/7 right?
[18:00:24 CEST] <drv> BBB: i have #ffmpeg and #ffmpeg-devel from early 2009 on
[18:00:25 CEST] <drv> https://drv.nu/irclogs/
[18:00:37 CEST] <BBB> \o/
[18:00:58 CEST] <BBB> now to find #libav logs
[18:01:06 CEST] <BBB> JEEB: if you have anything on #libav, thatd be awesome
[18:18:03 CEST] <BBB> drv: that goes until may 2014 btw
[18:18:14 CEST] <BBB> Ill use the mailinglist for the rest I suppose
[18:21:19 CEST] <drv> oh, i have the newer stuff too, just haven't merged the files yet
[18:21:23 CEST] <drv> i dropped them in the same dir
[18:29:50 CEST] <BBB> drv: ty!
[18:29:59 CEST] <BBB> now only to find libav logs
[18:30:04 CEST] <BBB> #libav logs, that is
[18:31:06 CEST] <JEEB> BBB: I do have some, but they are currently stranded between two different systems and at least three different directories
[18:31:17 CEST] <JEEB> so I would most probably be the best of sources
[18:31:31 CEST] <BBB> Ill take whatever I can get
[18:31:41 CEST] Action: BBB puts on his pretty please face
[18:32:54 CEST] <JEEB> I'll look into exact details soon (tm)
[18:33:05 CEST] <JEEB> but if I can scrape stuff off I can deliver it
[18:35:53 CEST] <BBB> ty! would be very much appreciated
[20:12:40 CEST] <c_14> Is there any reason av_freep would segfault when av_free doesn't? The segfault occurs in the call to free() according to the backtrace.
[20:13:21 CEST] <rcombs> c_14: forgot a &?
[20:13:46 CEST] <c_14> Ah, yep. Thanks.
[20:13:47 CEST] <c_14> That was stupid.
[20:15:44 CEST] <BBB> did you call it with &?
[20:15:44 CEST] <BBB> av_free(ptr) vs. av_freep(&ptr)
[20:16:02 CEST] <rcombs> BBB: beat you to it :)
[20:16:50 CEST] <BBB> I think my internet is shitty
[20:18:45 CEST] <BBB> like, I swear, in my chat window, its like as if I said it first
[20:18:59 CEST] <rcombs> then one of us has some significant lag
[20:19:21 CEST] <BBB> [2:13pm] <BBB> did you call it with &?
[20:19:22 CEST] <BBB> [2:13pm] <BBB> av_free(ptr) vs. av_freep(&ptr)
[20:19:22 CEST] <BBB> [2:15pm] <rcombs> c_14: forgot a &?
[20:19:22 CEST] <BBB> [2:15pm] <c_14> Ah, yep. Thanks.
[20:20:22 CEST] <rcombs> yeah, happens when ~latency~
[20:20:57 CEST] <BBB> probably me
[20:20:57 CEST] <BBB> my internet is shit today
[20:21:17 CEST] <kurosu> more like the irc server you're on?
[20:21:28 CEST] <kurosu> there's a solid 2 minutes between rcombs and you
[20:27:09 CEST] <BBB> who knows
[21:13:44 CEST] <wm4> remind me, how do I force the video decoder for ffprobe?
[21:14:10 CEST] <nevcairiel> -c:v no worky?
[21:14:14 CEST] <wm4> no
[21:31:35 CEST] <wm4> yeah, seems not possible
[23:06:55 CEST] <cone-899> ffmpeg 03Michael Niedermayer 07master:005d0069389a: avcodec/libopenh264enc: Avoid usage of deprecated ff_alloc_packet()
[23:06:55 CEST] <cone-899> ffmpeg 03Michael Niedermayer 07master:7d3aa2ca63c2: avutil/float_dsp: Remove use of deprecated av_set_cpu_flags_mask()
[23:33:46 CEST] <wm4> trying to use a debugger with ffmpeg is always such a drag
[23:34:03 CEST] <wm4> you can't disable optimizing because of some retarded "cute hack" someone made
[23:34:15 CEST] <wm4> installing strips by default, even if compiled with -g (????)
[23:35:17 CEST] <wm4> and setting "STRIP=" in config.mak somehow segfaults here
[23:36:22 CEST] <Shiz> STRIP=echo
[23:36:27 CEST] <Shiz> ( V?V)
[23:36:31 CEST] <wm4> yes that works
[23:36:33 CEST] <wm4> and it's lame
[23:36:49 CEST] <jamrial> did you try --disable-stripping during configure?
[23:37:02 CEST] <wm4> no, who thinks of this
[23:37:19 CEST] <Daemon404> dont forget --optflags=-O0
[23:38:50 CEST] <jamrial> wouldn't -O0 choke with all the stuff that depends on dead code elimination?
[23:39:02 CEST] <jamrial> --disable-optimizations sets it at -O1
[23:40:11 CEST] <iive> try -Og
[23:40:44 CEST] <iive> i haven't tested it, but somebody should try it :)
[23:44:39 CEST] <jamrial> wm4: try running configure with --enable-debug=gdb --disable-optimizations --disable-stripping for a gdb debug build
[23:47:13 CEST] <wm4> usually I just add #pragma GCC optimize("O0")
[23:47:19 CEST] <wm4> I was annoyed by the crappy defaults
[23:47:26 CEST] <wm4> why strip when it's a debug build?
[23:47:38 CEST] <wm4> or whatever --enable-debug is supposed to mean
[23:50:33 CEST] <BBB> I usually just compile the one file I care about with O0
[23:50:38 CEST] <BBB> the rest stays in O3
[00:00:00 CEST] --- Sat Aug  8 2015

More information about the Ffmpeg-devel-irc mailing list