[Ffmpeg-devel-irc] ffmpeg-devel.log.20140403

[burek]

[00:40] <cone-863> ffmpeg.git 03Senthilnathan M 07master:96b8e7a9cd3a: hlsenc: added floating point time support.
[00:40] <cone-863> ffmpeg.git 03Michael Niedermayer 07master:f359bac27cf2: avformat/hlsenc: remove unneeded l prefix from %lf
[01:44] <cone-863> ffmpeg.git 03Calvin Walton 07master:08909fb56b4f: Add Win32 GDI-based screen grabbing
[01:45] <cone-863> ffmpeg.git 03Michael Niedermayer 07master:8a0c16448c0e: MAINTAINERS: add gdi maintainer
[01:45] <cone-863> ffmpeg.git 03Michael Niedermayer 07master:a0258153730a: MAINTAINER: Add CC for dshow maintainer
[03:45] <cone-863> ffmpeg.git 03Michael Niedermayer 07master:fd2bcfc4d696: avcodec/jpeglsdec: fix palette parsing code
[13:19] <superware> what does it mean if av_write_frame returns -22?
[13:21] <nevcairiel> 22 is typically invalid argument or invalid value
[13:21] <nevcairiel> it doesn't mean anything by itself, look at the log output, it usually tells you which value was wrong
[13:31] <cone-109> ffmpeg.git 03Roman Fietze 07master:e12a73246d8c: dict: const pointers to dictionary where possible
[13:48] <cone-109> ffmpeg.git 03Roman Fietze 07master:74aeb6b58472: libavformat/utils.c: av_pkt_dump_log2: use const pointers
[14:05] <superware> nevcairiel: thanks
[14:22] <cone-109> ffmpeg.git 03Roman Fietze 07master:f5d718ac7e20: avfilter.h, buffer.c: use const src pointer in avfilter_copy_buffer_ref_props
[14:58] <ubitux> wm4: "WARNING: library configuration mismatch"
[14:58] <wm4> ubitux: yes?
[14:59] <ubitux> maybe we could abort if that's incompatible
[14:59] <ubitux> instead of just warn
[14:59] <wm4> how about... don't let the library user do that
[14:59] <ubitux> abort all the time?
[15:00] <wm4> abort if the libs are incompatible
[15:00] <ubitux> yeah what i suggested
[15:00] <ubitux> so... only if major mismatch maybe?
[15:01] <wm4> whatever ABI compatibility demands
[15:01] <ubitux> that won't work with libav
[15:01] <ubitux> we'll need some kind of compat array or something
[15:02] <ubitux> anyway, it should probably be another ticket
[15:03] <wm4> so when does ABI compatibility break?
[15:03] <ubitux> only between major versions?
[15:03] <ubitux> + libav conditional
[15:03] <wm4> then IMO it shouldn't crash
[15:04] <ubitux> major versions match? (didn't look closely)
[15:04] <wm4> (but ok, I see here it was swapped with an older version, and the approach the ffmpeg API takes makes it impossible to handle this properly)
[15:04] <wm4> yes
[15:04] <ubitux> same major, both ffmpeg?
[15:04] <wm4> yes
[15:04] <ubitux> ah if it's backward too it won't work
[15:05] <ubitux> obviously
[15:05] <wm4> yes, but IMO it's still some major bullshit (which user would expect this?)
[15:05] <ubitux> ?
[15:06] <ubitux> it means apps, ffmpeg included, will work with >= versions of the library
[15:06] <ubitux> which means they don't need to be rebuild
[15:06] <ubitux> what's the problem here?
[15:07] <wm4> I don't quite recall how this works on the dynamic linker level, but IMO it should either work, or fail to start at all
[15:10] <wm4> so maybe the soname should be bumped on every ABI enhancing change?
[15:10] <wm4> I'm saying "enhancing" because there are changes which break forward compatibility, but not backwards compatibility
[15:11] <wm4> like adding a field to a public struct
[15:43] <ubitux> i finally got my avx2 hw...
[15:44] <nevcairiel> that took quite a while
[15:44] <ubitux> yeah well
[15:44] <ubitux> fuck lenovo
[17:05] <cone-109> ffmpeg.git 03Lukasz Marek 07master:eaed4da96ac6: lavu/opt: extend AVOptionRange by extra values
[17:05] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:09656707cc75: Merge commit 'lukaszmluki/master^'
[17:57] <J_Darnley> Have you guys seen the email on ffmpeg-user titled "[FFmpeg-user] countless dropping frema warnings"?
[17:58] <Daemon404> frema? is that like the crema of frames?
[17:58] <Daemon404> if frames were like espresso.
[17:58] <J_Darnley> about ffmpeg spamming messages when dropping frames
[17:58] <Daemon404> ffmpeg cli doesnt spam duplicate messages
[17:58] <Daemon404> it gathers them
[17:58] <Daemon404> lets see
[17:59] <J_Darnley> It does what he says
[17:59] <Daemon404> ah ok theyre not dupes per se
[17:59] <J_Darnley> oh that's true
[17:59] <J_Darnley> but they still shouldn't be printed
[17:59] <Daemon404> my personal opinion is that if you are doing somethign as destructive as droppign frames
[18:00] <Daemon404> you should be lodu about it by default
[18:00] <Daemon404> silently droppign frames is evil
[18:00] <J_Darnley> This is too loud
[18:00] <Daemon404> i would disagree
[18:00] <Daemon404> they could maybe be bunched up into one warning
[18:01] <Daemon404> and then a sum at the end
[18:01] <Daemon404> it definitely needs to be that spammy at loglevel verbose though
[18:01] <Daemon404> which could be chanegd
[18:02] <J_Darnley> That I agree with.  Verbose can spit out a message about every frame
[18:02] <Daemon404> i'd wait for others opinions though
[18:02] <Daemon404> i am hardly authorative
[18:03] <Daemon404> lol
[18:03] <Daemon404> people *use* the cinepack encoder?
[18:03] <Daemon404> i don't believe that.
[18:03] <J_Darnley> :)
[19:16] <cone-109> ffmpeg.git 03Lukasz Marek 07master:8c2729237368: lavu/opt: dont crash on freeing NULL AVOptionRanges
[19:30] <cone-109> ffmpeg.git 03Luca Barbato 07master:006c253346d9: doc: Fit the graph in 80 cols
[19:30] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:0768fee9649a: Merge commit '006c253346d90f4abaf69d78083c8ba031628598'
[19:30] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:4169d8f6014e: doc/ffmpeg: restore ascii arrows in Simple filtergraphs
[20:53] <cone-109> ffmpeg.git 03Luca Barbato 07master:7763118cae4e: log: Support for 256color terminals
[20:53] <cone-109> ffmpeg.git 03Luca Barbato 07master:061e340c05bd: log: Factorize check_color_terminal() out
[20:53] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:c16e00618122: Merge commit '7763118cae4eb468b032dbd29af15a011c2c233b'
[21:08] <cone-109> ffmpeg.git 03Luca Barbato 07master:28b9ac4e48f3: log: Remove undefined behaviour
[21:08] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:002feaf6e22c: Merge commit '28b9ac4e48f3405a82e8e87ead336188215cf1e6'
[21:16] <cone-109> ffmpeg.git 03Luca Barbato 07master:42ba1c51d9e8: log: Document AV_LOG_C usage
[21:16] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:d9ac00ca7a8c: Merge commit '42ba1c51d9e843135772a3b9e5cf036a37b28bdb'
[22:23] <cone-109> ffmpeg.git 03Luca Barbato 07master:4ede1862dc62: log: Do not assume the TERM variable is always set
[22:23] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:5a18b215f713: Merge remote-tracking branch 'qatar/master'
[22:27] <J_Darnley> Oh great!  Why do we now have 256 colour support?
[22:31] <wm4> it's for debugging
[22:31] <wm4> why it's useful for debugging? no clue
[22:32] <wm4> or maybe it's just for "bugging"
[22:32] <J_Darnley> That's more like it
[22:32] <J_Darnley> I was about to send a patch to restore colours on Cygwin but now it doesn't work correctly
[22:43] <J_Darnley> and I think that's because some idiot hasn't tested on a not 256 colour terminal!
[22:54] <J_Darnley> Yes, as I thought, it doesn't check correctly for 256 colours
[22:54] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:1f8a6bef7828: avutil/log: add code to test av_log()
[22:54] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:49e040e86803: avutil/log: fix negative log levels
[22:56] <michaelni> J_Darnley, it will e fixed in a moment
[22:58] <J_Darnley> If you can see the problem, good because I can't at present
[23:00] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:609ef89703db: avutil/log: remove stray ;
[23:02] <J_Darnley> :o
[23:06] <J_Darnley> Much better
[23:07] <J_Darnley> I'm glad you saw that semi-colon.
[23:07] <J_Darnley> I wonder if there is a warning flag for that...
[23:12] <michaelni> if there is one we should definitly enable it
[23:13] <J_Darnley> I couldn't see one on my first pass though `man gcc` looking for "colon"
[23:21] <J_Darnley> How can there not be a warning option for that?  Perhaps I just can't see it.
[23:23] <Plorkyeran_> -Wempty-body
[23:24] Action: J_Darnley wonders how he missed that
[23:41] <J_Darnley> Do I need to use the American "color" in my commit messages?
[23:42] <kierank> no
[23:43] <kierank> perhaps in libav
[23:43] <J_Darnley> Oh I've got one "color" anyway in the title line anyway
[23:54] <J_Darnley> Comments on this, particularly the commit message: https://gitorious.org/ffmpeg/jdarnley-ffmpeg/commit/283e408d6c760ae0a799c22dc2599326ee411c5d
[00:00] --- Fri Apr  4 2014