[Ffmpeg-devel-irc] ffmpeg-devel.log.20120923

[burek]

[00:00] <ohsix> i must not be adjacent to those people
[00:00] <ohsix> punching comes after a few minutes of discussion too
[00:00] <Daemon404> to be fair i probably deserved it -- i troleld fairly hard
[00:00] <Daemon404> i compared rms to hilter
[00:00] <Daemon404> hitler*
[00:01] <ubitux> and he fell for that?
[00:01] <Daemon404> ubitux, 'passionate' people
[00:02] <ohsix> heh who would feel personally responsible for rms, he doesn't make excuses for himself or need any help :]
[00:03] <Daemon404> there's a small section of people who think he's a profet
[00:03] <Daemon404> er
[00:03] <Daemon404> prophet
[00:03] <ubitux> he actually has some quite accurate statements about the situation
[00:03] <ohsix> i wouldn't go anywhere those people are :[
[00:03] <ubitux> if you omit the sensible and repetitive 4-freedoms talk
[00:04] <ohsix> behead those who insult fsf
[00:04] <Daemon404> i love the fsf
[00:04] <Daemon404> they are and endless source of amusment
[00:04] <Daemon404> with their religious condemnations and preaching
[00:05] <ubitux> TBH, if you have to compare RMS to the two other steves, i wouldn't say RMS is the worse...
[00:05] <Daemon404> who's teh 2nd steve?
[00:05] <ubitux> jobs & ballmer
[00:06] <Daemon404> oh. sweaty blad man who cant run a company.
[00:06] <Daemon404> bald even
[00:07] <ubitux> so we have: 1) a useless commercial deadman 2) a monkey 3) and someone eating his feet and saying the same things for years, but with interesting ideas and philosophy
[00:07] <ubitux> choose your poison
[00:07] <Daemon404> i dont agree with 1)
[00:08] <Daemon404> nor 3)
[00:08] <ubitux> oh and you have hitler too, if you don't like conformism
[00:08] <ubitux> Daemon404: so you would choose ballmer?
[00:08] <Daemon404> to be clear, i dont actually think it was a valid comparison
[00:08] <Daemon404> purely for trolling
[00:09] <ubitux> it could have been interesting
[00:09] <ubitux> assuming you are going to the end of your reflexion
[00:09] <ubitux> like saying both are a demonstration of faschism in some sense
[00:09] <ubitux> a godwin point doesn't mean the end of the debate, it's kind of the start of an interesting one
[00:11] Action: Daemon404 notes irc has made him unproductive today.. oh well, saturday
[00:11] <ubitux> Daemon404: saturday has always been the most unproductive day in the development history of ffmpeg
[00:12] <ubitux> i have proof.
[00:12] <ubitux> oh or maybe it was monday
[00:12] Action: ubitux is going to reload the stats
[00:16] <ubitux> Daemon404: http://lucy.pkh.me/ffstats/activity.html#day_of_week
[00:16] <ubitux> 1 commit less than monday, i was right!
[00:16] <Daemon404> lul
[00:17] <ubitux> Daemon404: the good thing is, tomorrow you will be very productive
[00:17] <ubitux> according to the graph
[00:17] <ubitux> and stats never lie
[00:19] <saste_> is it the banner by herve flores?
[00:19] <saste_> i was fond of the September logo
[00:20] <ubitux> the font makes me think that yes it's from herve flores
[00:21] <saste_> i suppose there was no contest then?
[00:21] <ubitux> dunno
[00:21] <saste_> no contest as no contributed logos
[00:21] <ubitux> i remember a pretty awesome logo too
[00:21] <saste_> apart the ones from herve
[00:21] <ubitux> but i can't find it anymore
[00:22] <saste_> there is a trac ticket with more herve's designs
[00:22] <ubitux> oh that one: http://www.movieconverter-studio.com/_PUBLIC/ffmpeg/logo-new/periodical_logos/med/09-september.jpg
[00:22] <ubitux> this is the one i loved :)
[00:22] <saste_> me too, that was the one for september :)
[00:23] <ubitux> the current one is too as well though
[00:23] <ubitux> s/too/nice/
[00:24] <beastd> hi
[00:25] <ubitux> hey beastd 
[00:25] <ubitux> beastd: still willing to review webvtt?
[00:25] <ubitux> you have a few hours left
[00:25] <ubitux> i'm gonna sleep soon, and if i see no comment when i wake up, i'll push it
[00:26] <beastd> ubitux: ok. when will you wake up ;)
[00:27] <ubitux> you have to be aware, i'm sleeping 20 minutes in a night
[00:29] <beastd> that's tough
[00:32] <ubitux> :)
[00:32] <beastd> ubitux: I am already a bit tired. I helped a relative moving into a new appertment today. So ATM I am not feeling like reviewing code. But that can change.
[00:33] <ubitux> i'll push in ~12 hours
[00:33] <ubitux> but anyway, it's pretty independant, code can be fixed later if necessary
[00:34] <ubitux> beastd: if you feel like reviewing my other patches too... ;)
[00:38] <saste_> ubitux: tomorrow i'll review EBUR
[00:38] <saste_> ebu r128
[00:38] <saste_> seems in pretty good shape though
[00:39] <ubitux> no hurry for ebur128, but review is very welcome
[00:40] <ubitux> there are still a lot of work to do, but it's ready for a first version
[00:40] <ubitux> it would be pretty nice to see metadata injection and a normalization filter now ;)
[00:40] <Daemon404> wouldnt any normalization filter suck
[00:40] <Daemon404> it has to be 2pass to be any good
[00:43] <ubitux> i don't know yet how we are supposed to normalize with the loudness value, but i'd say a volume filter would be enough for the adjustment
[00:44] <ubitux> but i really didn't look how that's supposed to be done yet
[00:46] <JEEB> "normalization" should be 2pass, if it's just a dumb volume filter that can be on runtime
[00:47] <ubitux> ebur128 provides some moving loudness values, 400ms and 3s
[00:47] <ubitux> i guess this delay could be use for almost real-time
[00:48] <JEEB> there's something similar in ffdshow-tryouts and it sucks :|
[00:48] <ubitux> though, it might be wise to use the integrated value (whole track)
[00:48] <ubitux> JEEB: ebu r128 is the current loudness standard, it's pretty efficient afaict
[00:48] <Daemon404> anything sane writes does 2 pass, writes the metadata to the container
[00:49] <Daemon404> and it is used on playback
[00:49] <JEEB> well, all that I know about normalization is that all of the runtime stuff sucks diques
[00:49] <Daemon404> everything else is subpart
[00:49] <Daemon404> -t
[00:49] <JEEB> if you pardon my french
[00:49] <JEEB> because it has exactly that one area it is looking, and adjusts to it, possibly with a some kind of delay
[00:51] <burek> are there any interested developers to create an official static build of ffmpeg?
[00:52] <burek> or there is no interest for such thing
[00:52] <ubitux> mateo` can help you with that
[00:52] <ubitux> but you'll have to wait for next week
[00:52] <Daemon404> you would have to use uclibc
[00:52] <Daemon404> or similar
[00:53] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r7fe1ecefe1 10ffmpeg/libavfilter/vf_hue.c: 
[00:53] <CIA-56> ffmpeg: lavfi/hue: simplify/fix setting logic in set_options()
[00:53] <CIA-56> ffmpeg: Parse expression only when a new value is explicitly specified.
[00:53] <CIA-56> ffmpeg: In particular, avoid double free in case an old expression value is
[00:53] <CIA-56> ffmpeg: cached, it is set in the context, it is freed as the old value, and
[00:53] <CIA-56> ffmpeg: finally the pointer stored in the context is freed again when the filter
[00:53] <CIA-56> ffmpeg: is released.
[00:53] <CIA-56> ffmpeg: 03Stefano Sabatini 07master * r740c995288 10ffmpeg/ (6 files in 3 dirs): lavfi: add asendcmd and sendcmd filters
[00:53] <ubitux> \o/
[00:54] <saste_> now we need to add more commands to the existing filters
[00:54] <ubitux> yep
[00:54] <saste_> for example audio fade-in/out could be implemented through the volume filter
[00:54] <ubitux> hehe
[00:55] <saste_> video brightness fade-in/fade-out through an eq filter
[00:55] <ubitux> ffmpeg is so much fun @_@
[00:55] <ubitux> what's the state of the eq filter?
[00:55] <ubitux> nyuhu: are you working on it?
[00:56] <saste_> i ported it, but i'm confused about how i should handle eq/eq2
[00:56] <saste_> my wild guess is that eq2 is a subset of eq, so it would make sense to only support/port that
[00:56] <saste_> eq2 is a *superset of eq
[00:56] <ubitux>        eq[=brightness:contrast] (OBSOLETE)
[00:56] <ubitux>        eq2[=gamma:contrast:brightness:saturation:rg:gg:bg:weight]
[00:57] <ubitux> "OBSOLETE" sounds pretty obvious to me :p
[00:58] <saste_> yes, also i'm not the best person to port ASM code
[00:59] <saste_> and I'm feverly waiting for the patch from nicolas, which should simplify handling of silly positional mplayer arguments 
[00:59] <ubitux> i see only a small function
[01:01] <saste_> yes the port should be pretty straightforward
[01:01] <ubitux> and it could be omitted for a first version :p
[01:01] <ubitux> it's a special case
[01:04] <saste_> good night
[01:28] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * ra380f82c53 10ffmpeg/libavformat/utils.c: 
[01:28] <CIA-56> ffmpeg: lavf: allow using request_probe as a score threshold to accept a codec
[01:28] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:28] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r697b476c07 10ffmpeg/libavformat/wav.c: 
[01:28] <CIA-56> ffmpeg: wav: Fix DTS detection
[01:28] <CIA-56> ffmpeg: Fixes Ticket1474
[01:28] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:28] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * ra9f522d1b7 10ffmpeg/libavformat/avformat.h: 
[01:28] <CIA-56> ffmpeg: lavf: document request_probe properly
[01:28] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:59] <CIA-56> ffmpeg: 03rogerdpack 07master * r641c0eea0b 10ffmpeg/libavformat/riff.c: 
[02:59] <CIA-56> ffmpeg: bitmap header lookup: add code comment
[02:59] <CIA-56> ffmpeg: Signed-off-by: rogerdpack <rogerpack2005 at gmail.com>
[02:59] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:59] <CIA-56> ffmpeg: 03Duncan Salerno 07master * r0a8cc1179a 10ffmpeg/libavformat/http.c: 
[02:59] <CIA-56> ffmpeg: http: larger buffer for long URIs
[02:59] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:59] <CIA-56> ffmpeg: 03Duncan Salerno 07master * r48cf0f1c18 10ffmpeg/libavformat/http.c: 
[02:59] <CIA-56> ffmpeg: http: only send range header when necessary
[02:59] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:46] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r61a84755a9 10ffmpeg/libavformat/utils.c: 
[05:46] <CIA-56> ffmpeg: lavf: dont open a decoder at the top of find_stream_info before probing has finished
[05:46] <CIA-56> ffmpeg: It could open the wrong decoder.
[05:46] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:46] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r21c4393f1c 10ffmpeg/libavformat/utils.c: 
[05:46] <CIA-56> ffmpeg: lavf/find_stream_info: make sure we decode at least one frame for DTS
[05:46] <CIA-56> ffmpeg: The channel layout at the demuxer layer is commonly wrong and would
[05:46] <CIA-56> ffmpeg: otherwise be used.
[05:46] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:52] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rf0bb88e2bc 10ffmpeg/libavformat/http.c: 
[13:52] <CIA-56> ffmpeg: Revert "http: only send range header when necessary"
[13:52] <CIA-56> ffmpeg: This reverts commit 48cf0f1c18f2f91363601c94177c71bc3ac13b8e.
[13:52] <CIA-56> ffmpeg: This can cause problems with seekability detection
[13:58] <CIA-56> ffmpeg: 03Justin Ruggles 07master * r0ccf051a9d 10ffmpeg/libavcodec/utils.c: 
[13:58] <CIA-56> ffmpeg: avcodec: fix memleak in avcodec_encode_audio2()
[13:58] <CIA-56> ffmpeg: Ensure that padded_frame is freed before returning when needed.
[13:58] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rfecfdf6e72 10ffmpeg/: 
[13:58] <CIA-56> ffmpeg: Merge remote-tracking branch 'qatar/master'
[13:58] <CIA-56> ffmpeg: * qatar/master:
[13:58] <CIA-56> ffmpeg:  avcodec: fix memleak in avcodec_encode_audio2()
[13:58] <CIA-56> ffmpeg: Conflicts:
[13:58] <CIA-56> ffmpeg:  libavcodec/utils.c
[13:58] <CIA-56> ffmpeg: Merged-by: Michael Niedermayer <michaelni at gmx.at>
[14:25] <CIA-56> ffmpeg: 03Dmitry Samonenko 07master * r697ea4fccf 10ffmpeg/libavformat/ (Makefile rtpdec.c rtpdec_formats.h rtpdec_speex.c): (log message trimmed)
[14:25] <CIA-56> ffmpeg: Introducing speex RTP demuxing (RFC 5574)
[14:25] <CIA-56> ffmpeg: RTPDynamicProtocolHandler for speex is added. Initial support for
[14:25] <CIA-56> ffmpeg: speex depacketization from RTP stream comes with it.
[14:25] <CIA-56> ffmpeg: Currently, only codec audio rate can be applied based on sdp:
[14:25] <CIA-56> ffmpeg: * Narrowband ( 8K)
[14:25] <CIA-56> ffmpeg: * Wideband (16K)
[14:25] <CIA-56> ffmpeg: 03Dmitry Samonenko 07master * rb024c41e1e 10ffmpeg/libavformat/ (rtpenc.c sdp.c): 
[14:25] <CIA-56> ffmpeg: Add initial support for speex RTP packetization (RFC 5574)
[14:25] <CIA-56> ffmpeg: * libspeex audio codec is no longer considered unsupported
[14:25] <CIA-56> ffmpeg:  when using rtp as output format.
[14:25] <CIA-56> ffmpeg: * SDP rtpmap is added for speex payload, formatted according to RFC
[14:25] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:18] <CIA-56> ffmpeg: 03Clément BSsch 07master * r2599541b3d 10ffmpeg/ (14 files in 6 dirs): WebVTT demuxer and decoder.
[15:40] <CIA-56> ffmpeg: 03Reimar Döffinger 07master * r3905caf251 10ffmpeg/libavcodec/mpeg12.c: 
[15:40] <CIA-56> ffmpeg: Document that the MPEG12 "fast" functions can read out of bounds.
[15:40] <CIA-56> ffmpeg: This is intentional (and a reason why fast usually should not be used),
[15:40] <CIA-56> ffmpeg: but a few people sent patches to "fix" this recently so it would be
[15:40] <CIA-56> ffmpeg: better to document this.
[15:40] <CIA-56> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[15:40] <CIA-56> ffmpeg: 03Reimar Döffinger 07master * r033f53fb8f 10ffmpeg/libavformat/http.c: 
[15:40] <CIA-56> ffmpeg: Document why we always send a Range HTTP header.
[15:40] <CIA-56> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[17:38] <CIA-56> ffmpeg: 03Sebastien Zwickert 07master * rf25d53d95f 10ffmpeg/configure: 
[17:38] <CIA-56> ffmpeg: configure: disable VDA on ppc, fix build
[17:38] <CIA-56> ffmpeg: Fixes Ticket1760
[17:38] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:38] <ro-ee> I recently upgraded my entire system, and when recompiling mplayer, in ffmpeg I get the following error: AR	libavfilter/libavfilter.a
[18:38] <ro-ee> ar: libavfilter/allfilters.o: File format not recognized....     What could be missing? Did make distclean and everthing, but thus far no luck.
[21:01] <Daemon404> saste_, ffprobe problem just got a lot more evil
[21:06] <CIA-56> ffmpeg: 03Nicolas George 07master * r7200fecd9b 10ffmpeg/cmdutils.c: 
[21:06] <CIA-56> ffmpeg: cmdutils: sort codec lists.
[21:06] <CIA-56> ffmpeg: The lists are sorted first by type (video first)
[21:06] <CIA-56> ffmpeg: and then alphabetically by name.
[21:06] <CIA-56> ffmpeg: Fix ticket #1669.
[21:08] <Daemon404> ubitux, i might have to add a very basic webvtt encoder... :<
[21:08] <Daemon404> against my will
[21:09] <ubitux> haha
[21:09] <ubitux> for you company? :)
[21:10] <ubitux> +r
[21:10] <Daemon404> indeed
[21:10] <ubitux> fun
[21:10] <Daemon404> easy way to convert sub format X into webvtt
[21:10] <ubitux> writing an encoder is more painful than a decoder in the current state
[21:10] <ubitux> have fun
[21:10] <Daemon404> better than a bunch of perl scripts
[21:10] <ubitux> parsing ASS ;)
[21:10] <Daemon404> or well
[21:11] <Daemon404> maybe i'll use libaegisub
[21:11] <Daemon404> xD
[21:15] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r127b70e423 10ffmpeg/libavcodec/ra144.c: 
[21:15] <CIA-56> ffmpeg: ra144: factor division out
[21:15] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:19] <Daemon404> saste_, native mingw ffprobe tests fail!
[21:19] <Daemon404> inb4 works via wine
[21:20] <Daemon404> -            "comment2": "I e Üñîçød¬"
[21:20] <Daemon404> +            "comment2": "I â"¥ ÃSñîçødâ¬"
[21:20] <Daemon404> btw the latter is a vlaid utf8 string representing this.
[21:20] <Daemon404> valid*
[21:21] <Daemon404> i wonder if this relies on the shell.
[21:32] <ubitux> if this unicode string is that much trouble we could just get rid of it
[21:32] <ubitux> though, i'd like to test a bit the escaping
[21:33] <ohsix> there are no unicode strings, just strings with one of many unicode encoding \m/
[21:33] <ubitux> so maybe something a string with \x01\x02\x03 would be nice
[21:35] <Daemon404> i
[21:35] <Daemon404> have a hunch
[21:35] <Daemon404> and i will deskface if im right
[21:35] <ohsix> you should get that looked at
[21:35] <Daemon404> lol
[21:42] <Daemon404> ............
[21:42] <Daemon404> me deskfaces
[21:42] <Daemon404> AHHHH
[22:18] <Daemon404> https://japland.org/tools/linux/ffprobe_nut_ref_creation.html
[22:18] <Daemon404> ^ problem lies in the ref file creation on windows
[22:18] <Daemon404> i think it might have to do with passing utf8 cmd line args in windows
[22:18] <ubitux> fun
[22:20] <Daemon404> oops those filenames are backwards
[22:22] <Daemon404> F5
[22:23] <Daemon404> "Unfortunately, the encoding used by the Windows command prompt is never(*) UTF-8. It's a locale-specific encoding, so you can only pass Japanese characters in an argument on a Japanese Windows install."
[22:23] <Daemon404> fuck yeah
[22:25] <ubitux> :D
[22:25] <Daemon404> is there anyway to read metadata tags to use from a file?
[22:25] <ubitux> metadata foobar="$(cat file)" ?
[22:25] <Daemon404> that still passes it the same
[22:26] <ubitux> what about hardcoding the binary values?
[22:26] <Daemon404> as a cmd line arg
[22:26] <ubitux> $(printf "\x...\x...")
[22:26] <Daemon404> lol
[22:26] <Daemon404> thats feasible maybe
[22:26] <ubitux> :)
[22:26] <Daemon404> lemme try i guess
[22:28] <Daemon404> is printf even POSIX?
[22:29] <Daemon404> as in bourne shell
[22:29] <ubitux> should be way more portable than echo at least
[22:29] <ubitux> from a usage PoV
[22:29] <ubitux> not sure about the availability
[22:30] <Daemon404> $ printf "\xEF\xBB\xBF\x49\x20\xE2\x99\xA5\x20\xC3\x9C\xC3\xB1\xC3\xAE\xC3\xA7\xC3\xB8\x64\xE2\x82\xAC"
[22:30] <Daemon404> I e Üñîçød¬
[22:30] <Daemon404> well lets try
[22:31] <ubitux> :)
[22:31] <pengvado> http://pubs.opengroup.org/onlinepubs/009695399/utilities/printf.html
[22:31] <ubitux> Daemon404: careful, you're including the UTF-8 bom
[22:31] <Daemon404> oh
[22:31] <Daemon404> woops
[22:31] <Daemon404> so i am
[22:31] Action: Daemon404 removes
[22:43] <Daemon404> ubitux, nah pritn didnt do shit
[22:44] <Daemon404> printf*
[22:44] <Daemon404> same results as passing it normally
[22:45] <ubitux> haha
[22:45] <ubitux> too bad :)
[22:49] <ubitux> Daemon404: and directly the string as such?
[22:49] <ubitux> (without the printf)
[22:50] <Daemon404> yes
[22:50] <Daemon404> same behavior
[22:50] <Daemon404> it IS getting passed correct info
[22:50] <Daemon404> windows cmd line shit mangles it
[22:50] <ubitux> nice software you have there
[22:51] <ubitux> oups sorry hate etc ;)
[22:51] <Daemon404> msdn wants you to use wmain, which whats wchar_t sorts of stuff
[22:51] <Daemon404> and thats just retarded
[22:51] <ubitux> so you can't pass arbitrary binary data to your program from the cmd line?
[22:52] <Daemon404> no
[22:52] <Daemon404> no you cannot
[22:52] <Daemon404> nor should you imho
[22:52] <ubitux> why not?
[22:52] <Daemon404> cmdline args are not for passing binary data
[22:52] <Daemon404> its incredibly stupid
[22:55] <ubitux> and so there is no solution?
[22:55] <Daemon404> we cannot load metadata from a file?
[22:56] <Daemon404> its possible, but i dont wanna rape ffprobe code with this wmain bs
[22:56] <Daemon404> just so we can do a fate test with a unicode cmd line arg
[22:56] <ubitux> isn't it possible to reverse the operation within ffmpeg?
[22:56] <Daemon404> we could also use CommandLineToArgvW
[22:56] <ubitux> basically charset convert if WIN32
[22:56] <Daemon404> i think
[22:57] <Daemon404> but it uses a wchar_t still
[22:57] <ubitux> http://stackoverflow.com/questions/388490/unicode-characters-in-windows-command-line-how
[22:57] <ubitux> did you see that?
[22:57] <ubitux> "chcp 65001" sounds magic
[22:57] <Daemon404> its wrong
[22:57] <Daemon404> and not valid for msys
[22:57] <ubitux> :(
[22:58] <Daemon404> dont believe everythign you read on teh internet
[22:58] <Daemon404> im talking to my ms buddy
[22:58] <Daemon404> there might be a nonitristive way
[22:58] <ubitux> are you insulting him?
[22:58] <Daemon404> ms has nothing to do with insulting him
[22:58] <Daemon404> i insult him anyway
[22:59] <ubitux> :D
[22:59] <ubitux> yup, that has always proved to be more efficient 
[22:59] <ubitux> (bad wording but you get the point)
[23:00] <Daemon404> indeed
[23:16] <Daemon404> not working so well
[23:18] Action: Daemon404 pings saste_ 
[23:32] <Daemon404> ubitux, it is far worse... there is some nonstandard behvior in MS's runtime
[23:32] <Daemon404> that breaks fucking common sense and logic
[23:33] <ubitux> :D
[23:33] Action: Daemon404 is investigating
[23:37] <saste_> Daemon404, pong
[23:38] <saste_> ah the wtf16 stuff
[23:38] <Daemon404> nah
[23:38] <Daemon404> its not utf816
[23:38] <Daemon404> its some nonstandard behavior with argv
[23:38] <Daemon404> on windows
[23:40] <saste_> yep the usual charencoding winmess fun
[23:42] <saste_> ok so we can't set a unicode string with -metadata when using windows cmdline
[23:44] <Daemon404> yea
[23:44] <Daemon404> let me investigate a bit more
[23:44] <saste_> so basically it is not ffprobe to fail, but ffmpeg when setting the value
[23:44] <Daemon404> yes
[23:45] <Daemon404> it might be in nutenc.c
[23:45] <ubitux> what's wrong in trying to reverse the operation within ffmpeg internals?
[23:45] <saste_> well we can workaround that, if the goal is to get the test pass
[23:45] <Daemon404> cause i dumped the contents of argv
[23:45] <Daemon404> and theure utf8
[23:50] <Daemon404> saste_, we seem to be doing fun stuff like sending utf8 shit through toupper
[23:50] <Daemon404> i dunno whats going on :|
[23:50] <saste_> Daemon404, where?
[23:51] <saste_> toupper is used only for section names
[23:51] <saste_> should not be used for anything else
[23:51] <Daemon404> i see
[23:51] <Daemon404> well tracing how exactly metadata is written from nutenc.c
[23:51] <Daemon404> is friggen hard
[23:51] <Daemon404> all this avdict crap
[23:52] <saste_> are you sure the string you read from commandline is proper UTF-8?
[23:52] <Daemon404> i dumped teh bytes
[23:52] <Daemon404> its byte-for-byte that utf8 string
[23:52] <saste_> ok that's weird
[23:52] <saste_> did you try to use a metadata file?
[23:53] <saste_> i can provide an example if you want
[23:53] <Daemon404> example owuld be nice
[23:53] <Daemon404> ill try it if so
[23:53] Action: saste_ reads the docs
[23:53] <Daemon404> lol
[23:54] <Daemon404> i didnt think we could read them
[23:54] <saste_> no we are not allowed indeed
[23:56] <Compn> oh no not windows utf nonsense
[23:56] <Compn> ignore ignore!
[23:57] Action: Compn dir /x and uses 8.3 filenames
[23:57] <Daemon404> saste_, aha!
[23:57] <Daemon404> av_dict_set uses av_strdup
[23:57] <Daemon404> which use strlen
[23:57] <Daemon404> time to dump infoz from it
[23:57] <ubitux> and strlen doesn't work on windows?
[23:58] <Daemon404> it might work differently with utf8 strings
[23:58] <ubitux> that would be really weird.
[23:58] <saste_> from what i remember, strlen() should return the lenght in bytes, rather than the lenght in characters
[23:58] <ubitux> it would have break in a lot of other places
[23:58] <Daemon404> it should yes
[23:58] <saste_> for our purposes this should be correct
[23:58] <saste_> but then i won't bet that windows strlen is behaving properly
[00:00] --- Mon Sep 24 2012