[Ffmpeg-devel-irc] ffmpeg-devel.log.20140901

burek burek021 at gmail.com
Tue Sep 2 02:05:02 CEST 2014


[00:13] <llogan> do most people use a mail client that can parse attachments (such as *.patch) as text? or do they download/save and view w/ external tool?
[00:16] <Timothy_Gu> llogan: I don't use a mail client, so the latter.
[00:16] <kierank> latter
[00:17] <jamrial> thunderbird shows attached text files just fine here, regardless of extension
[00:18] <wm4> claws-mail (the worst mail client in existence) can show them inline just fine
[00:18] <wm4> there's even a very unintuitive way to quote it for reply
[00:23] <J_Darnley> I use thunderbird too and sometimes attached patches are not shown following the message.  I guess this due to the "wong" mime type.
[00:23] <J_Darnley> *wrong
[00:24] <llogan> sounds like attached patches are annoying then
[00:24] <llogan> wm4: why use claws if it is the worst?
[00:25] <wm4> llogan: the alternatives are not very attractive either
[00:25] <llogan> yeah, i'm using it too
[00:25] <wm4> claws has a collection of hilarious bugs though
[00:25] <J_Darnley> I'm not too fussed but only because I am not a very good/active reviewer
[00:25] <llogan> wm4: pointed some out to the devs. seemed like nice dudes if not a little goofy.
[00:26] <llogan> J_Darnley: don't worry, i'm not a good patch writer.
[00:26] <J_Darnley> poor michael
[00:26] <llogan> I used to use alpine then mutt but not so much anymore
[00:27] <wm4> llogan: I basically got "patch welcome"
[00:28] <llogan> does thunderbird still dump all messages into a huge monolithic file? i forgot the format.
[00:29] <J_Darnley> Yes!
[00:29] <wm4> when I look in the thunderbird profile, I see many files
[00:29] <wm4> sqlite, dat, rdf, json...
[00:29] <J_Darnley> Doubly so if you let it grab gmail's "all mails" folder
[00:29] <wm4> pure crack
[00:29] <wm4> ini, xml, db
[00:32] <pross-au_> <- mutt
[00:34] <llogan> That Reindl guy in -user is an angry one.
[00:35] <wm4> btw. I've noticed that the gif demuxer doesn't report the file duration
[00:36] <wm4> is it unable to?
[00:36] <llogan> hmm...you're right. i'm not sure.
[00:40] <cone-745> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:b627385b4bd2: ffserver: tests, use image2 instead of pgmyuv
[00:53] <J_Darnley> I wonder if thunderbird got compression for its mail files.
[00:54] <J_Darnley> Somehow, my entore profile is "only" 367MB
[00:54] <J_Darnley> *entire
[00:55] <BtbN> "files"
[00:55] <BtbN> It's one huge plain mbox file
[00:55] <BtbN> With just all mails right next to each other
[00:55] <J_Darnley> Actually, I think it is one file for every imap folder.
[00:56] <J_Darnley> Anyway, they are still very large
[00:59] <wm4> apparently thunderbird still uses this format: http://www.jwz.org/blog/2004/03/when-the-database-worms-eat-into-your-brain/
[01:03] <BtbN> mbox isn't realy a format. It's just all the mails right after each other in one file
[01:03] <BtbN> One of that per logical folder
[01:04] <BtbN> Plus one index db file, which has offsets for each mail into the file, so it doesn't have to be parsed completely all the time
[01:13] <J_Darnley> Doe anyone have a tip about how I can stop vim from adding tabs into makefiles?
[01:14] <BtbN> :set paste
[01:15] <BtbN> But what do you want to do? Don't makefiles depend on there beeing tabs?
[01:15] <J_Darnley> How does ffmpeg's makefiles work then?
[01:15] <BtbN> ?
[01:15] <BtbN> It's using tabs?
[03:25] <cone-745> ffmpeg.git 03Michael Niedermayer 07master:42507534977e: avformat/mpegts: favor DESC_types over probed codecs
[04:06] <jamrial> i'm getting a "fatal: read error: Connection reset by peer" when i try to fetch commits from the main repository
[04:07] <jamrial> three different OSes, two different connections. same thing in all of them
[04:17] <michaelni> jamrial, cant reproduce
[04:17] <michaelni> also main git is provided by videolan, i cant help if it doesnt work ...
[04:22] <jamrial> ok, was just mentioning it in case others were having the same problem
[05:31] <cone-745> ffmpeg.git 03Michael Niedermayer 07master:37c46743ee42: avcodec: fix missing doxygen comment marker
[05:31] <cone-745> ffmpeg.git 03Michael Niedermayer 07master:97b88096dcbe: Merge commit '37c46743ee42df1d6986ba3c996e3af5b95511f7'
[09:04] <cone-943> ffmpeg.git 03Clément BSsch 07master:36952786a5cc: build: use pkg-config to detect libmodplug
[10:18] <ubitux> i wonder if we could use parallel in ffmpeg configure
[10:19] <ubitux> the thing i'm afraid of is that there are various side effects with all the global variables
[10:19] <ubitux> no functions are using "local" which sucks hard
[10:19] <ubitux> but OTOH it doesn't seem standard :(
[10:49] <cone-943> ffmpeg.git 03Stefano Sabatini 07master:d4ec07dfe7db: lavu/avstring: check for overlong encodings in av_utf8_decode()
[11:03] <ubitux> saste: btw, saw that http://bjoern.hoehrmann.de/utf-8/decoder/dfa/ ?
[11:05] <saste> ubitux, i know a lot could be optimized in the decoder
[11:24] <Daemon404> ubitux, solution: critical sections using file locks
[11:24] Action: Daemon404 runs
[11:24] <ubitux> heh
[11:24] <ubitux> Daemon404: i'm more concerned about stuff like this: http://git.videolan.org/?p=ffmpeg.git;a=blob;f=configure;h=80f27484101a3e545af7ecdd3b80606d44f5cb57;hb=HEAD#l1200
[11:25] <ubitux> here, $pkg get changed in check_pkg_config
[11:25] <Daemon404> ah yes
[11:25] <ubitux> it's not obvious what happens when require_pkg_config can be called with "libfoobar >= 1.2.3"
[11:25] <Daemon404> i forgot
[11:26] <ubitux> and still prints "libfoobar not found"
[11:26] <Daemon404> shell only *looks* scoped
[11:26] <ubitux> it can be if you use the local prefix
[11:26] <Daemon404> im not sure thats POSIX
[11:26] <ubitux> yes, my point earlier :(
[11:29] <Daemon404> ;p
[11:36] <cone-943> ffmpeg.git 03Clément BSsch 07master:343f07e15c58: RELEASE_NOTES: mention libmodplug detection change
[14:49] <J_Darnley> What does an ampersand (&) mean in a function argument in c++ (or maybe in msvc)?
[14:50] <wm4> it's a reference... it's like passing a pointer, except the variable is implicitly dereferenced on access
[14:50] <wm4> so it behaves like a normal value and not a pointer
[14:51] <J_Darnley> in that case, would I need to change it to a normal pointer in regular C?
[14:51] <wm4> yes
[14:51] <J_Darnley> okay thanks
[14:51] <wm4> it's just syntactic sugar
[14:52] <wm4> (or syntactic diabetes)
[14:52] <J_Darnley> :)
[14:53] <J_Darnley> ooh... I might need some brackets as well here, or maybe just some spaces: *d*=0.96
[14:54] <nevcairiel> it should work like that, but it sure looks weird
[14:54] <Daemon404> i never understoof the point of references myself
[14:54] <nevcairiel> the way i usually use them is to avoid a copy of a struct
[14:54] <wm4> copying variables can be very expensive in C++
[14:55] <Daemon404> i mean
[14:55] <Daemon404> why not use a const pointer?
[14:55] <wm4> because pointers are EVIL
[14:55] <nevcairiel> makes using stuff more noisy
[14:55] <Daemon404> i dunno, less bad that reusing & for already possible functionality
[14:55] <Daemon404> syntax wise
[14:56] <wm4> IMO the ABI should just specify that large structs are passed by pointer (and the compiler has to take care to keep the semantics equivalent)
[14:56] <J_Darnley> Then you would have people arguing over "large"
[14:58] <wm4> the ABI makes many of such tradeoffs
[14:58] <wm4> e.g. how many registers should be callee-saved etc.
[15:02] <ubitux> the references in c++ are so evil&
[15:02] <Daemon404> lots of things are
[15:02] <ubitux> float x = 3; foobar(x);  you don't know if foobar is going to modify x unless you look at the prototypes
[15:03] <Daemon404> and strangely, i am writing c++ today
[15:03] <Daemon404> because JSON libs in C are shitty to use
[15:03] <wm4> Daemon404 has went to the dark side
[15:03] <ubitux> i don't understand how people manage to understand c++ code
[15:03] <Daemon404> traversing trees and shit... no thanks
[15:03] <wm4> I wonder when ffmpeg will be converted to C++
[15:03] <Daemon404> i'd rather just unmarshall it in one func call (int oan obj/hash table)
[15:03] <wm4> seems to happen to all C projects eventually
[15:24] <TimNich> wm4:  some seem  to stay C with a C++ wrapper for the odd bits that are easier (such as Daemon404's JSON library)
[15:25] <Daemon404> TimNich, thats kinda what i do yeah
[15:25] <Daemon404> its very rare i need a c++ wrapper though
[15:25] <Daemon404> json in c just sucks so much.
[15:25] <nevcairiel> anything highly string intensive sucks much in c
[15:25] <Daemon404> ;p
[15:26] <Daemon404> it's nice to do parse() -> obj["key"]
[15:26] <Daemon404> or w/e
[16:05] <michaelni> anyone has any oppinon about "Timothy Gu      ( 15K) [FFmpeg-devel] [PATCH] Use makeinfo to generate html doc for the new website" ?
[16:06] <michaelni> this patch will remove support for building html docs for several major platforms
[16:06] <michaelni> ubunto 12.04 and OSX being 2 affected AFAIK
[16:07] <plepere> is there a documentation page for all the ASM macros ? (SPLATW and all)
[16:09] <michaelni> i only know the source as "documentation"
[16:09] <michaelni> if there really is none, maybe starting one could make sense, maybe on our wiki or somewhere on x264
[16:10] <kierank> splat just means take a value and repeat it across a simd register
[16:11] <kierank> the advantage being that the macro "chooses" the best instruction depending on the instruction set
[16:11] <plepere> kierank, I know, but there are plenty of macros (butterfly, clip...etc)
[16:11] <kierank> clip is similar to av_clip
[16:11] <kierank> butterfly i've never used
[16:11] <plepere> being able to have a list of them would be nice
[16:12] <kierank> ask in #x264dev
[16:12] <kierank> and then paste them somewhere
[16:12] <plepere> hmmm
[16:12] <plepere> I'll maybe do that in october
[16:15] <ubitux> seems we need a musl instance
[16:29] <nevcairiel> Screw all those obscure weirdos
[16:29] <BBB_> plepere: I can explain a couple if you have specific questions
[16:29] <BBB_> plepere: theres no overall-complete documentation ATM, but asking around is generally considered good
[16:29] <nevcairiel> :F
[16:30] <plepere> BBB_, no, I'm fine. I was just wondering. I might do a wiki entry about it in October once I'm at my new job.
[16:30] <BBB_> cool
[16:46] <plepere> well, once I've found a place to stay. I have to move to the UK.
[16:55] <J_Darnley> michaelni: if someone is building new ffmpeg from source on those "old" platforms, I don't see why we can't require them to have "new" tools.
[16:55] <J_Darnley> but then I also like supporting old platforms
[17:46] <cone-435> ffmpeg.git 03Timothy Gu 07master:8d6ec6118698: general: Fix usage of @float
[18:56] <ubitux> http://david.li/filtering/
[18:57] <wm4> "Unfortunately, your browser does not appear to support the required features."
[18:57] <wm4> aw
[18:57] <Daemon404> works here
[18:57] <Daemon404> inb4 paranoid slashdot tinfoilhat with no js
[18:59] <thardin> I sure love browser wars
[18:59] <saste> ubitux, cool, we really miss fourier filtering
[18:59] <wm4> just using a browser that doesn't need several hundreds MB of RAM to display a blank page
[18:59] <ubitux> saste: we have a dct filter ;)
[19:01] <wm4> hm even in firefox it shows that message
[19:01] <ubitux> works fine in firefox here
[19:01] <Daemon404> here too
[19:12] <J_Darnley> If you visit that page without javascript you don't get the message
[19:13] <J_Darnley> but you also don't get much else.
[19:42] <cone-435> ffmpeg.git 03Reimar Döffinger 07master:2a00812d82c5: h261dec, ituh263dec: Avoid unnecessary -1 inside inner loop.
[19:42] <cone-435> ffmpeg.git 03Reimar Döffinger 07master:36c0043344fd: h261dec, ituh263dec: ensure CLOSE_READER is done on error paths, too.
[19:42] <cone-435> ffmpeg.git 03Reimar Döffinger 07master:e8d0b84e8a58: h261dec, ituh263dec: Move condition.
[20:01] <ubitux> here we go, the ffmpeg depreceted message got reverted to something not so obvious
[20:01] <ubitux> they just don't want to say it's been "deprecated in the Libav project"
[20:01] <ubitux> :D
[20:02] <ubitux> this is surreal haha
[20:03] <wm4> what a horrifying patch (aac fixed point decoder)
[20:03] <wm4> patches of several hundred KBs
[20:04] <ubitux> the previous ffmpeg/deprecated msg was fine, why the hell did they submit yet another crap like that ffs :(
[20:04] <wm4> ubitux: it was fine? I thought it was not fine
[20:05] <ubitux> i remember seeing "deprecated in the Libav project" at some point
[20:05] <ubitux> which was just fine
[20:05] <nevcairiel> wm4: why do you think it takes ages to apply any of them, this stuff has been coming to the ML for a year now or so
[20:10] <jamrial> most are big because of code being moved to new files (like templates). also new huge tables
[20:12] <wm4> I understand the reasons (or at least some), but it's still horrifying
[20:28] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:a2dbf379f01b: avcodec/dvbsubdec: make compute_edt user settable
[20:28] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:b01d2a5533a9: ffmpeg: Enable compute_edt only when the subtitles are used for a output stream instead of a filter
[20:41] <cone-435> ffmpeg.git 03Diego Biurrun 07master:4600a85eaa61: dxva2: Pass variable of correct type to IDirectXVideoDecoder_GetBuffer()
[20:41] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:052cd1ce7204: Merge commit '4600a85eaa6182e5a27464f6b9cae5a9ddbf3098'
[20:52] <cone-435> ffmpeg.git 03Diego Biurrun 07master:b8962d64cc71: dxva2: Clean up definition of _WIN32_WINNT
[20:52] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:38c4316fff74: Merge commit 'b8962d64cc71af241601bcab5c3fcdc5735ef4ae'
[21:01] <cone-435> ffmpeg.git 03Gabriel Dume 07master:63f800ca5f4f: aac_adtstoasc_bsf: Return proper error code
[21:01] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:f626eaa58e69: Merge commit '63f800ca5f4f6b38f3789a7edb03bfabe8bacdd2'
[21:33] <cone-435> ffmpeg.git 03Gabriel Dume 07master:e205fe6831fe: mpeg4videodec: Fix low_delay error message
[21:33] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:f9aadf1fc426: Merge commit 'e205fe6831fe1b5445f7827b5c916425620b4ab5'
[21:50] <cone-435> ffmpeg.git 03Thomas Volkert 07master:f8164f6728a3: Add missing changelog entry
[21:59] <cone-435> ffmpeg.git 03Mark Harris 07master:1b3a98f13790: doc/filters.texi: fix filter name in examples
[23:31] <cone-435> ffmpeg.git 03Carl Eugen Hoyos 07master:7cf1f0f4af70: Fix standalone compilation of the snow decoder.
[00:00] --- Tue Sep  2 2014


More information about the Ffmpeg-devel-irc mailing list