[FFmpeg-devel-irc] IRC log for 2010-03-25

irc at mansr.com irc at mansr.com
Fri Mar 26 01:00:42 CET 2010


[00:42:46] <CIA-24> ffmpeg: stefano * r22664 /trunk/libavdevice/oss_audio.c:
[00:42:46] <CIA-24> ffmpeg: Make audio_read_packet() return AVERROR_EOF rather than AVERROR(EOF) =
[00:42:46] <CIA-24> ffmpeg: AVERROR(-1) = 1.
[02:40:03] <BMintern> hello. i see someone recently updated libavfilter (changing vf_overlay.c)...
[02:40:19] <BMintern> when i try to run "./checkout.sh", the patch is failing
[02:41:02] <BMintern> any ideas?
[02:45:45] <Dark_Shikari> so, for those here
[02:45:58] <Dark_Shikari> I just got a mail from a company offering big $ for anyone who RE'd prores
[02:46:01] <Dark_Shikari> Who should I tell them to talk to?
[02:48:01] <peloverde> kostya?
[02:48:12] <Dark_Shikari> anyone else?
[02:48:12] <Dark_Shikari> BBB?
[02:48:36] <Dark_Shikari> in their words: "big fat contract"
[02:59:13] <Compn> Dark_Shikari : look who wrote on prores page in wiki ?
[02:59:40] <Compn> oh nm, no one re'd it yet
[03:00:02] <Compn> 'maxpol' ?
[03:00:07] <Compn> maxim :)
[03:00:21] <Compn> aka indeo 4/5 guy
[07:14:11] <CIA-24> ffmpeg: mstorsjo * r22665 /trunk/libavformat/ (sdp.c rdt.c utils.c internal.h): Add a lowercase parameter to ff_data_to_hex
[07:15:31] <CIA-24> ffmpeg: mstorsjo * r22666 /trunk/libavformat/utils.c: Reindent
[09:26:52] <astrange> do av_crop_uint8 and ff_cropTbl both need to exist?
[09:27:05] <kshishkov> yes
[09:27:21] <kshishkov> their intended use differs
[09:48:23] <DonDiego> http://wurstball.de/static/ircview/pictures/749cd15bf9d0254286148f468567b29e.jpg
[09:48:25] <DonDiego> rotfl
[09:49:48] <andoma> :)
[09:50:27] <av500> fefe ftw
[09:58:01] <thresh> oh, a new version.
[11:06:20] <KotH> do we have any embedded linux developers here (for comercial, customer specific development)?
[11:06:44] <merbzt1> ->mru
[11:07:19] <KotH> he isnt awake yet ;)
[11:07:44] <kshishkov> nevertheless
[11:08:18] <KotH> aren't there any other?
[11:08:27] * KotH thought we have 3-5 of them around ffmpeg
[11:08:32] <av500> depends on the price :)
[11:09:13] <kshishkov> maybe a clone of mru
[11:09:31] <av500> KotH: is the tender public?
[11:09:32] <kshishkov> and that __troll__ guy
[11:09:37] <KotH> av500: nope
[11:09:40] <av500> ah
[11:10:10] <KotH> well... background is, we have a _poential_ customer who wants an embedded linux system and needs it quickly
[11:10:29] <KotH> (series production within 9 months, no hardware available yet)
[11:10:39] <kshishkov> give him gentoo ;)
[11:10:48] <av500> KotH: sounds doable
[11:10:52] <KotH> and i'm looking for ways to bootstrap the development as fast as possible
[11:11:02] <av500> and the hw is?
[11:11:14] <av500> PIC or XEON?
[11:11:26] <KotH> av500: it's only doable if 1) we get the contract now, 2) and we have someone who knows how to build a embedded linux system from scratch
[11:11:44] <av500> lfs ftw
[11:11:53] <av500> KotH: what cpu?
[11:11:56] <KotH> av500: atm, the plan goes for an arm9 based system for linux and an arm7 for the realtime stuff
[11:12:11] <KotH> (to avoid the need of using a real time linux)
[11:12:29] <av500> depending an what you need to realtime, you dont need rt linux
[11:12:54] <KotH> sensor input to actuator motion within <10ms
[11:13:13] <av500> run it besides the kernel on the same arm
[11:13:31] <KotH> what do you mean?
[11:14:28] <av500> we did something like that for an IR reciever, linux irq latency was way to long to sample the IR code, so we hooked it to FIQ and linux kernel did not know anything about it
[11:14:59] <av500> it only got the decoded IR msg through some mailbox registers
[11:15:11] <KotH> you mean, you made a "driver" that hooked into FIQ and provided the kernel with the data?
[11:15:19] <av500> yes
[11:15:29] <av500> kernel was in IRQ and we were in FIW
[11:15:33] <av500> FIQ
[11:15:41] <av500> so we had priority
[11:15:48] <DonDiego> av500: you saw the article on fefe's blog? hilarious and oh so true...
[11:15:53] <av500> yes
[11:16:23] <KotH> av500: problem here: high sampling rate measurement -> lots of FIQs -> kills performance on the cpu
[11:16:38] <KotH> DonDiego: what article?
[11:16:52] <av500> http://blog.fefe.de/?ts=b5546b0c
[11:17:23] <KotH> i'd be glad if you just had an interrupt source that tells us when we have to switch, but it's actually based on sensor input with a software/runtime defined limit
[11:18:10] <av500> KotH: well, if you cannot preprocess the irq stuff to a more moderate level, then I guess you have to burn another CPU for it...
[11:18:27] <KotH> juup
[11:18:38] <KotH> well.. i first have to get more accurate specs from the customer
[11:18:51] <av500> anybody con do it with accurate specs :)
[11:19:10] <KotH> atm we have only "we want something that works within half a year"
[11:19:21] <av500> I am sure they will know what they want once you built it...
[11:20:15] <KotH> originally it was only about HW (processor board with sensor/actuator interface). but no requirements given but the most basic ones (10 sensors, 20 actuators,...), no exact spec how the sensors look like, no spec what the software should do,...
[11:20:16] <av500> for the arm9 side something like oe/angstrom can spare you the "from scratch" part
[11:21:20] <av500> maybe you should hire an "oracle"
[11:21:29] <KotH> an oracle?
[11:22:22] <av500> einen hellseher
[11:22:56] <KotH> nah.. i'm currently writing a long mail asking for various things
[11:23:10] <KotH> but my knowledge in the area of embedded linux is kind of limited
[11:23:17] <av500> it is like linux
[11:23:26] <KotH> so, i'm trying to guestimate things from my basic linux knowlege
[11:26:24] <av500> if you do all the RT stuff on the arm7, then the arm9 linux can be pretty standard..
[11:26:33] <KotH> yes
[11:26:44] <av500> but yes, detailed specs would help: )
[11:26:53] <KotH> though, i'm not worried about user space stuff. as soon as i have a kernel running i know how to handle it
[11:27:14] <KotH> the big question is whether we need to write any drivers for the hw and how much time that needs
[11:27:14] <av500> running linux on an arm9 is not rocket surgery any more these days
[11:27:27] <av500> I guess you need a driver to talk to the arm7
[11:27:41] <av500> unless you connect it with an uart :)
[11:27:49] <KotH> not if we can do it as an RS232 or so :)
[11:28:03] <av500> I2C :)
[11:28:04] <KotH> or i2c or spci...
[11:28:08] <KotH> lol
[11:28:10] <KotH> gtma :)
[11:28:20] <KotH> s/tm/mt/
[11:28:28] <av500> :)
[11:32:12] <twnqx> KotH: did you ever design with PCIe? :P
[11:33:42] <av500> twnqx: ah, that slow stuff
[11:34:05] <twnqx> slow stuff?
[11:34:07] <KotH> twnqx: nope
[11:34:10] <twnqx> i'm satisfied with it
[11:34:15] <twnqx> too bad
[11:34:35] <KotH> twnqx: we hardly ever need such fast connections, and if we need them, then pci-e is too slow
[11:34:52] <av500> see, slow! :P
[11:35:21] <kshishkov> pci-e keyboard?
[11:35:30] <twnqx> i'm looking at around ~150MB/s crypto throughput :P
[11:36:03] <kshishkov> firewire?
[11:36:11] <KotH> atm we are developing a system that uses up to twelve ccd chips and transferes the raw data over wires on an MID substrate -> no chance for pci-e with it's strict requirements
[11:36:30] <KotH> twnqx: private or comercial?
[11:37:03] <twnqx> first for me
[11:37:07] <twnqx> if it works, why not sell it
[11:38:45] <KotH> well.. you see... private, i have no time to play with stuff like that... but if you would hire us, i'd have enough time :)
[11:39:04] <KotH> but we aren't cheap
[11:39:19] <KotH> (in both senses of the word)
[11:40:43] <KotH> twnqx: btw: you dont need pci-e for 150-MByte/s
[11:40:54] <KotH> twnqx: 64bit PCI should be enough
[11:41:02] <av500> usb 3.0 ftw
[11:41:02] <twnqx> yeah, but i won't design for obsolete interfaces.
[11:41:11] <KotH> twnqx: if it isn't, go to 64bit at 66MHz
[11:41:17] <KotH> obsolete?
[11:41:25] <KotH> you'd be surprised :)
[11:41:27] <twnqx> yeah, i consider parallel PCI obsolete
[11:41:40] <twnqx> my laptop has only mini-pcie and expresscard left
[11:41:42] <kshishkov> next thing somebody will call RS232 obsolete
[11:41:45] * KotH was designing stuff for PCMCIA not even two years ago
[11:41:47] <twnqx> my htpc only mini-pcie
[11:42:00] <twnqx> my desktop has loads of PCIe, but just one PCI
[11:42:34] <twnqx> kshishkov: luckily even router vendors realise the death of rs232
[11:42:53] <twnqx> cisco e.g. implements an ACM TTY as usb client these days :P
[11:44:06] <KotH> lol
[11:44:18] <KotH> w/o rs232, the whole electronics development would be dead
[11:44:39] <av500> KotH: they just slap on FTDIs everywhere
[11:44:47] <av500> the cpus themselves still have uarts...
[11:45:27] <twnqx> no, not FTDIs, but something similar
[11:45:35] <twnqx> FTDIs do proper usb serial ports :P
[11:45:42] <twnqx> not generic acm tty
[11:46:09] <kshishkov> twnqx: usb = universal parallel bus, isn't it?
[11:46:33] <KotH> av500: yeah..
[11:46:41] <av500> kshishkov: UnStable Bus
[11:46:43] * KotH is glad his olimex programmer has an serial interface
[11:46:50] <KotH> av500: stop using windows
[11:46:59] <twnqx> KotH: what do you need the serial for, except for communicating with legacy devices?
[11:47:00] * av500 needs only to look at his FTDIs to make them lock up
[11:47:02] <KotH> av500: usb is working very well on linux
[11:47:14] <av500> KotH: I know
[11:47:18] <KotH> twnqx: uhm.. developing the software for the devices we build?
[11:47:33] <twnqx> and you don't flash via JTAG/SPI?
[11:47:37] <kshishkov> av500: I still prefer FireWire
[11:47:40] <KotH> twnqx: lol
[11:47:50] <twnqx> or use ethernet for communication?
[11:47:53] <twnqx> :S
[11:48:07] <KotH> twnqx: jtag is good for programming. even proper debuggin is a pain on most systems
[11:48:22] <KotH> twnqx: being able to use an rs232 for debug output is a great help
[11:48:25] <av500> KotH: you cant ssh into your PIC?
[11:48:32] <KotH> nah..
[11:48:44] <KotH> not enough ram to hold the data needed for the tcp connection
[11:49:03] <twnqx> which reminds me that i'm still pondering whether to buy that xilinx EDK or not...
[11:50:29] <twnqx> *sigh*
[11:50:42] <kshishkov> twnqx: may I point out that some interfaces may be slow since interfaced device is slow. Human, for example
[11:50:55] <twnqx> this is not about speed
[11:51:01] <twnqx> jtag/spi are slow too
[11:51:43] <twnqx> it's more about usability, USB is a de-facto standard these days, RS232 is gone. i consider adding one of these "tiny usb" controllers to my designs
[11:52:04] <twnqx> basically, an 8pin avr doing a software implementation of 12mbit USB :P
[11:52:17] <kshishkov> yuck!
[11:52:34] <av500> twnqx: thats were the FTDIs are nice, you have GIOs on usb...
[12:25:55] <twnqx> av500: i have an fpga... just looking at usb endpoint implementations :P
[12:28:11] <av500> twnqx: trivial, no? :)
[12:28:44] <twnqx> almost :P
[12:28:57] <twnqx> i think my choice of a 48MHz oscillator wasn't too bad...
[12:36:29] <KotH> lol
[12:36:43] <KotH> twnqx: how do you want to implement ch9 on the fgpa?
[12:37:05] <twnqx> ch9?
[12:38:12] <KotH> chapter 9
[12:38:20] <KotH> ie chapter 9 of the usb specs
[12:38:49] <KotH> or do you use the fpga only as hw interface to a cpu ?
[12:38:56] <twnqx> nah
[12:39:09] <twnqx> (unless i buy EDK and go microblaze)
[12:39:53] <KotH> so, you want to implement an usb device completely within the fpga?
[12:39:58] <twnqx> yeah
[12:40:05] <KotH> that'll be fun :-)
[12:40:14] <twnqx> uh
[12:40:25] <twnqx> opencores.org has usb cores...
[12:40:37] <KotH> and what do those usb cores do?
[12:41:11] <KotH> or to give you some perspective: i've been working on and off the last 1.5y on an usb framework for teh atmel sam7 uC's
[12:41:15] <KotH> and i'm not finished yet
[12:42:19] <KotH> if you know usb well, it'll be probably 2-3 man months to implement a usb framework and get a simple class like a com port working
[12:42:23] <kshishkov> why "uC's" sounds like "really small capacitors"?
[12:42:46] <KotH> kshishkov: that's uF... or rather pF ;)
[12:43:19] <kshishkov> capacitors are marked as "C" on circuit schemes here
[12:43:38] <kshishkov> and "LC" is not for "low-complexity" either
[12:44:12] <twnqx> no, a coil-capacitor-filter :P
[12:44:38] <KotH> kshishkov: uC = \mu C in non-ascii writing = micro controller
[12:45:00] <kshishkov> I suspected that
[12:45:18] <twnqx> hm.
[12:45:24] <KotH> kshishkov: maybe you should learn some electronics :)
[12:45:49] <kshishkov> oh, I did
[12:45:56] <av500> kshishkov: you can RE electronics too :)
[12:46:07] <KotH> let me guess, everything was written in kyrilic?
[12:46:42] <kshishkov> almost
[12:47:10] <kshishkov> we still use Latin letters for electronic part names
[12:49:45] <kshishkov> I don't think we've made past op-amps at electronics
[12:50:32] <kshishkov> though we also another couse in electronic design  and we had to make CPU model for project on that
[12:50:57] <KotH> eh..
[12:50:58] <kshishkov> *course
[12:51:05] <KotH> you know that you cannot learn electronics design at school?
[12:51:13] <KotH> unless you have a very very good school...
[12:51:15] <twnqx> depends
[12:51:37] <kshishkov> well, even "very good school" does not apply here
[12:52:06] <twnqx> but yeah, unless you get your hands on the stuff during school... you're in a bad shape indeed
[12:52:16] <KotH> twnqx: even if you do
[12:52:19] <twnqx> and few schools have the financial resources to do that
[12:52:25] <KotH> twnqx: most schools are just too short to teach you the art
[12:52:43] <KotH> twnqx: they can give you the basics, but to become half way decent, you have to teach it yourself
[12:53:08] <twnqx> well, if i'd ever need more of those small FPGA boards than i have now, i'd make some changes
[12:53:22] * KotH was at a uni that had the financial resources to let students build their own asic, but still didnt learn much about electronics design there
[12:53:32] <kshishkov> KotH: in our case it was not intended even for bootstrapping
[12:53:49] <KotH> kshishkov: it? bootstrap what?
[12:54:00] <kshishkov> more like horrible subject you need to pass to be called real student
[12:54:01] <twnqx> what's there to learn? PCB layout for HF, the "black magic"?
[12:54:14] <twnqx> layout for analog?
[12:54:22] <kshishkov> KotH: yourself. With knowledge obtained at school.
[12:54:49] <KotH> twnqx: even basic analog design
[12:55:02] <KotH> twnqx: like, how to design a single transistor amplifier
[12:55:03] <kshishkov> you should know - "here are basics, there are books, go and learn"
[12:55:11] <twnqx> well yeah, i'm currently failing at that :P
[12:55:39] <KotH> the furthest we got, was that one prof mentioned a miller capacitor
[12:56:15] <twnqx> and i don't even want something like a linear amplifier. i just want a 2.5signal to turn on a 5V signal.
[12:56:38] <twnqx> and i feel like that's only possible with a dual inverter setup.
[12:56:49] <KotH> huh?
[12:56:55] <kshishkov> huh? Schmidt trigger ftw
[12:57:03] <KotH> bc548 and be done with it
[12:57:18] <twnqx> meh, according to spice that doesn't work
[12:57:22] <KotH> possible with a comparator if you need a level detection
[12:57:26] <twnqx> nah
[12:57:49] <twnqx> hm 548
[12:57:56] <KotH> uhm... if you use spice, you have to know what you simulate
[12:58:03] <KotH> otherwise you'll get a lot of crap output
[12:58:11] <KotH> spice is a typical GIGO system
[12:58:17] <twnqx> true that
[12:58:52] <KotH> and even if you used the right input, chances are high that the simulation results are still crap
[12:59:54] <twnqx> looking at the first 548 schematic... that's exactly what i used to 547s for >_>
[12:59:58] <twnqx> two*
[13:00:27] <twnqx> i mean, i know why i failed at the first attempts, once i remembered the operations of transistors :P
[13:00:37] <KotH> 547s and 548s are nearly the same transistor anyways :)
[13:00:56] * twnqx looks at the spice schematic again
[13:00:57] <KotH> and both are debrecated ;)
[13:01:05] <twnqx> sure, but still available
[13:01:09] <KotH> use 847 and 848 instead :)
[13:01:22] <kshishkov> guess what country still produces vacuum lamps ;)
[13:01:37] <KotH> ofc, because of all the tinkerers who dont know any better
[13:02:49] <twnqx> hm
[13:03:04] <twnqx> 847 is not available in leaded... just SMD
[13:03:13] <kshishkov> no, there's a special class of idiots who thinks lamps are better and want to pay any sum of money for them
[13:04:20] <KotH> twnqx: lol
[13:04:32] <KotH> twnqx: nobody uses leaded stuf for small fry like that anymore ;)
[13:04:44] <twnqx> if i'd go make a PCB - no
[13:05:15] <KotH> kshishkov: actually, there are places where normal lamps are better
[13:05:28] <twnqx> like.. everywhere?
[13:05:47] * twnqx dislikes the idea of more toxic things to accidently break around him
[13:05:49] <KotH> twnqx: making a pcb is cheap these days
[13:06:12] <twnqx> it's not the 2€ i pay for a copper strips board unless i make it myself
[13:06:13] <kshishkov> KotH: of course, wanna get a delivery?
[13:06:29] <twnqx> and there i'm too lazy
[13:07:23] <KotH> twnqx: if you have just handfull of discrete components, that's enough
[13:07:31] <twnqx> also, i'm still using breadboards :P
[13:07:35] <KotH> twnqx: for anything else, i'd make a pcb
[13:07:39] <twnqx> yeah
[13:07:42] <twnqx> agreed :)
[13:07:46] <KotH> breadboards? with fpgas?
[13:07:52] <twnqx> lol no
[13:08:03] * KotH knows people who'd do that
[13:08:03] <twnqx> though my PCB could be used to plug that in :]
[13:08:34] <twnqx> it's not for no reason that the 40 I/O pins exactly fit in a standard 2.54mm socket..
[13:11:42] <KotH> http://en.wikipedia.org/wiki/Inversion_of_control <- could someone explain me what's written there? i dont understand anything but "we want an abstraction"
[13:12:30] <kshishkov> let me see, I know many buzzwords
[13:15:02] <kshishkov> sounds like usual management here - task is simply passed from highest level to lowest worker unchanged who has to make something out of it and comply
[13:15:08] <mru> KotH: I see java and patterns mentioned
[13:15:11] <mru> ergo, it's rubbish
[13:15:20] <mru> and factory
[13:17:07] <KotH> for me it sounded like the usual way of defining APIs for lower level stuff
[13:17:19] <mru> oh god, they've invented something so abstract they don't even know what it is
[13:17:24] <KotH> and layering everything
[13:17:25] <kshishkov> someone should ban use of word "patterns" in such cases so those CS theorists will go in search of real work
[13:17:29] <mru> "It is still undecided if Inversion of Control is a design pattern, ..."
[13:18:01] <kshishkov> KotH: not layering alone but delegating
[13:22:10] <KotH> kshishkov: hmm.. for me layering implies delegation
[13:22:21] <KotH> if you dont delegate, then layering doesnt make sense
[13:22:26] <kshishkov> KotH: http://img-fotki.yandex.ru/get/3309/turboblog.0/0_50c7_99d05bc4_XL.jpg - here's only low layer has any control on everything :)
[13:22:29] <KotH> unless i misunderstand what delegation means
[13:22:54] <kshishkov> IoC means delegation without decision
[13:23:22] <mru> the step from ioc to ioccc ain't big...
[13:23:35] <KotH> kshishkov: that pic is originally german ;)
[13:23:42] <kshishkov> you just push whole task down and at that level it decides what part of that task was intended for it
[13:24:03] <KotH> oh..
[13:24:07] <KotH> now i get it
[13:24:09] <KotH> ok...
[13:24:14] <kshishkov> KotH: not necessarily. It's quite Russian in spirit
[13:24:17] <KotH> mru: you're wrong
[13:24:20] <KotH> mru: it's not crap
[13:24:25] <KotH> mru: it's _CRAP_
[13:24:29] <kshishkov> ioccc is art
[13:25:33] <mru> is ioc that style where you can be searching for a week without finding the code that _actually does anything_
[13:26:04] <kshishkov> but you can be sure that it's not in the first files you see
[13:27:21] <kshishkov> reminds me of aspect-oriented programming (aka wishful thinking)
[13:28:15] <mru> http://daringfireball.net/2010/03/on_submarine_patents
[13:28:19] <mru> kshishkov: isn't that haskell?
[13:28:49] <kshishkov> I've heard about that from Java dev
[13:29:42] <mru> for all it flaws, there are actually big, working apps written in java
[13:29:48] <mru> you can't say that about haskell
[13:30:17] <jai> xmonad
[13:30:19] * jai hides
[13:30:40] * kshishkov remembers his prof saying that programming is not an art anymore because of those development strategies
[13:32:50] <kshishkov> I'm glad they have not mentioned patterns at all during CS courses
[13:33:21] <twnqx> pattern aren't bad
[13:33:31] <kshishkov> they are buzzword
[13:33:36] <twnqx> it's like my first algorithm and data structures lessons
[13:33:47] <twnqx> "oh, THAT'S the name for what i invented for myself"
[13:34:05] <twnqx> it's a neat way to use common names for things
[13:34:07] <mru> patterns are good for describing things
[13:34:10] <mru> not for designing thing
[13:34:11] <mru> s
[13:34:23] <KotH> i'd disagree
[13:34:28] <twnqx> many things you design folow the same patterns
[13:34:32] <mru> yes
[13:34:32] <twnqx> follow*
[13:34:34] <kshishkov> you'll get philosophy next
[13:34:48] <mru> but that doesn't mean you should force every design into one of a few pre-defined patterns
[13:34:51] <twnqx> forcing everything into a limited set of pattern obviously fails
[13:34:56] <twnqx> :]
[13:34:58] <mru> and that's what they do
[13:35:00] <KotH> it's good to know a bunch of patterns you can choose from, but not to design by trying to use a certain pattern and make it fit to your prob
[13:35:15] <kshishkov> twnqx: no, that depends on your power
[13:35:32] <twnqx> heh
[13:35:58] <KotH> kshishkov: coding has nothing to do with dragon balls!
[13:36:08] <mru> you design using a _strategy_, and a pattern will emerge
[13:36:14] <mru> the opposite does not happen
[13:36:34] <KotH> knowing patterns help to find your strategy :)
[13:36:41] <mru> maybe
[13:36:53] <KotH> .o0(when we've finished bow, we should write a book on programming)
[13:36:55] <mru> but again, that is not what they teach
[13:36:59] <KotH> nope
[13:37:10] <KotH> actually, they dont teach programming at all
[13:37:25] <KotH> you're in a "good" uni if they teach you patterns
[13:37:29] <mru> they did in a few classes I took
[13:37:36] <mru> but they were not taught by CS dept
[13:37:45] <KotH> YOU'VE BEEN TAINTED!!
[13:37:49] <KotH> ;)
[13:38:18] <mru> they made us write things like memory allocators and task schedulers
[13:38:31] <mru> and pipelined instruction scheduling
[13:38:53] <twnqx> KotH: back on the 548... it would invert the single, wouldn't it?
[13:39:04] <twnqx> signal*
[13:39:19] <twnqx> not that that would be an issue
[13:39:47] <twnqx> hm well... on the other hand, it would be... i need 5V or nothing, not 5V or GND
[13:40:13] <mru> nothing?
[13:40:17] <mru> Z-state?
[13:40:23] <kshishkov> KotH: local joke - "military IQ tests (put object of certain shape into a hole with the same shape) have divided soldiers into two groups - very dumb and very strong"
[13:40:36] <twnqx> high impedance would be a working term, yes
[13:41:22] <kshishkov> so just use threshold cutoff on buffer control port ;)
[13:41:22] <twnqx> i need to be able to turn off some pullup resistorsto stop influencing the line
[13:41:47] <mru> look for parts with tristate outputs
[13:41:56] <mru> what kind of parts are we talking about here?
[13:42:09] <twnqx> the output of some level shifters :P
[13:42:24] <mru> tristate level shifters... never heard of that
[13:42:31] <twnqx> oh, they are
[13:42:39] <twnqx> but it's kind of a loopback test
[13:43:15] <twnqx> with some lines driving a LCD at the same time - the LCD need 5V, so there are pullups
[13:43:25] <twnqx> for the loopback tests the pullups need to be disabled
[13:51:55] <BBB> my connection is apparently not that great
[13:53:00] <kshishkov> use dialup ;)
[13:53:22] <kshishkov> and maybe it was bad idea to go via IPv6
[13:53:42] <BBB> ?
[13:53:45] <BBB> I'm on ipv6?
[13:53:48] * BBB had no idea
[13:53:49] <kshishkov> you was
[13:53:58] <kshishkov> 15:41 -!- BBB [~BBB at 2002:8cfb:2314:8:216:cbff:fe8d:def0] has joined #ffmpeg-devel
[13:54:13] <BBB> awesome! \o/
[13:54:21] <kshishkov> now it shows your usual uni address
[13:54:22] <BBB> but it doesn't work, so screw ipv6
[13:54:44] <merbzt1> yeah who needs that
[13:54:51] <kshishkov> CISCO?
[13:54:54] <merbzt1> ipv4 4tw
[13:54:57] <wbs> BBB: well, in 10 years perhaps ipv6 will be mature and ready to be used ;-)
[13:54:59] <BBB> superdump: could you reply to marcelo's email on the list?
[13:55:33] <BBB> if cisco wants to sell, maybe they should create products that A) do not infringe on everyone's GPL copyrights, and B) products that are actually novel, innovative and high-quality
[13:55:46] <BBB> was it cisco that bought kiss?
[13:56:09] <BBB> yes it was, linksys and thus cisco
[13:56:46] <superdump> which list?
[13:56:52] <BBB> ffmpeg-devel@
[13:57:06] <BBB> Marcelo Galvao Povoa
[13:57:09] <BBB> wants to do amrwb decoder
[13:57:13] <BBB> you're suggested mentor for that
[13:57:15] <BBB> I think
[13:57:18] <superdump> aha
[13:57:22] <superdump> yes, i am
[13:57:39] <BBB> he's asking for a qual task
[13:57:50] <BBB> he sounds like he might be suited for the job
[13:58:58] <superdump> mmm
[13:59:01] <superdump> so i see
[13:59:19] <CIA-24> ffmpeg: mstorsjo * r22667 /trunk/libavformat/ (httpauth.h httpauth.c): Add support for http digest authentication
[13:59:31] <KotH> twnqx: use a 558 and high-side switching
[13:59:49] <BBB> wbs: sorry about the method/uri thing, but I dislike the splitting a bit
[14:00:23] <BBB> and yay for digest auth
[14:00:26] <BBB> I've wanted that for long
[14:00:30] <BBB> add a changelog entry please
[14:00:39] <wbs> ah, yeah. minor bump, too?
[14:00:45] <BBB> don't know
[14:00:57] <wbs> BBB: and no problem with the method/uri thing, i agree that it's fugly
[14:01:01] <KotH> twnqx: and while you are at it, buy the tietze-schenk
[14:01:06] <KotH> twnqx: you need it
[14:01:12] <twnqx> nah
[14:01:13] <BBB> koth: did you install the cms?
[14:01:19] <KotH> BBB: no time yet
[14:01:19] <twnqx> i don't do analog deisgn usually :X
[14:01:22] <BBB> ok
[14:01:31] <superdump> gah tb3 crashed while i was responding
[14:01:36] * KotH is fighting against customers and an s6 code he wrote in a hurry
[14:01:49] <KotH> BBB: i hope i can do it on sat/sun
[14:01:58] <KotH> BBB: no guarranties though
[14:02:14] <KotH> BBB: you can try to pester mru or diego too ;-)
[14:02:30] <CIA-24> ffmpeg: mstorsjo * r22668 /trunk/Changelog: Add a changelog entry for http digest auth
[14:02:33] <wbs> BBB: do you prefer keeping the current approach with specifying a full "cmd" buffer, including the method/uri, or should the method/uri be factored out from the cmd buffer and only provided separately?
[14:03:29] <BBB> KotH: ok
[14:04:22] <BBB> wbs: I think we already copy into a new buffer in that ff_rtsp_send_cmd_with_content_async() function, so I'd say just provide method/uri separately and have send_cmd_..() construct the full thing
[14:04:35] <BBB> it sounds a little more sane also
[14:04:52] <wbs> BBB: ok.. will give it a try, but it may take until tomorrow :-)
[14:05:13] <BBB> that's ok
[14:05:24] <BBB> then I can review some other patches in the mean time ;)
[14:05:29] <wbs> exactly ;-)
[14:06:10] <BBB> or, god forbid, actually code myself :)
[14:11:43] <BBB> wbs: can you confirm that with that patch, the binary does not change md5?
[14:12:13] * BBB wonders if the compiler will put a terminating 0 in some random unallocated memory location
[14:14:14] <wbs> BBB: it does change here, but I think it should be ok anyway. perhaps with making the length 17 (or just []) would be even safer
[14:14:32] <BBB> I'm hoping to prevent wasting that extra byte ;)
[14:14:47] <wbs> ah. ;-)
[14:14:48] <BBB> I'm wondering how it changes
[14:35:09] <CIA-24> ffmpeg: jai_menon * r22669 /trunk/libavformat/matroskaenc.c: Matroska muxer : Don't create audio tracks for unsupported audio codecs.
[14:35:52] * KotH curses at customers
[14:36:01] <kshishkov> :)
[14:36:17] <KotH> why are they silent for most of the year and the resurface all at the same time with something very urgent?
[14:36:20] <justlooking> KotH> and what do those usb cores do? < http://www.opencores.org/project,usbhostslave apparently it still needs a little work though...
[14:38:38] <KotH> justlooking: that's chapter 7 and part of chapter 8, no chapter 9
[14:38:45] <kshishkov> KotH: in my experience what customers have is either urgent or can be neglected
[14:39:18] <KotH> kshishkov: ack
[14:39:32] <KotH> justlooking: ie, you still need some controller who controls the endpoings
[14:39:38] <KotH> endpoints
[14:40:00] <KotH> justlooking: and there is still no class done if you have chapter 9...
[14:40:25] <KotH> justlooking: ie, what you call "little work" is quite a bit of work
[14:40:52] <twnqx> justlooking: it also has a wishbone interface, so i expect it needs some CPU
[14:41:24] <twnqx> but i'd need to actually read chapter 9 to fully judge :]
[14:41:51] <KotH> twnqx: reading is no problem, understanding all implication though is one ;)
[14:42:09] <twnqx> bah, it can be dirty and limited
[14:42:16] <twnqx> it just needs to work somehow â„¢
[14:42:21] <KotH> twnqx: if you want to do some usb stuff, get yourself a copy of "usb complete" by axelson
[14:42:31] <KotH> lol
[14:42:55] <KotH> usb isnt rs232, you cannot just make a dirty hack and hope it'll work
[14:42:56] <twnqx> right now i'm trying to figure the difference between bc558, bc558a, bc558b and bc558c >_>
[14:43:02] <KotH> especially not if you try to interface windows
[14:43:09] <twnqx> well, you know the tiny i2c?
[14:43:14] <twnqx> ah, winblows
[14:43:17] <twnqx> no, i won't
[14:43:22] <twnqx> i refuse to program on it
[14:43:34] <KotH> interfacing with linux is easier, at least linux tells you what it doesnt like about the device
[14:43:44] <KotH> twnqx: you'll need windows anyways...
[14:44:00] <KotH> twnqx: download the usb compliance tester code from intel
[14:44:02] <twnqx> to jump out after trying to understand the usb spec?
[14:44:15] <KotH> twnqx: it runs only on windows, but tells you a lot about your device and what it is doing wrong
[14:44:35] <twnqx> hm
[14:44:41] <KotH> oh.. and get a beagle 480 too... you'll need it ;)
[14:45:10] <twnqx> http://www.harbaum.org/till/i2c_tiny_usb/index.shtml <- run against that
[14:45:19] <justlooking>  there are others there too, search on "usb" there and you see several.
[14:46:28] <KotH> twnqx: looks nice
[14:46:39] <KotH> twnqx: if you can use one of those instead of building your own usb device... use that
[14:46:57] <KotH> twnqx: even interfacing an ftdi with high speed serial is simpler
[14:47:00] <KotH> and prolly cheaper
[14:47:09] <KotH> usb is a fucking beast of a protocol
[14:47:16] <twnqx> it's based around some GPL usb stack
[14:47:19] <twnqx> with no endpoint
[14:47:29] <KotH> rotfl
[14:47:35] <KotH> every usb device has at least one endpoint
[14:47:46] <KotH> otherwise it is no usb device ;)
[14:47:47] <twnqx> http://www.obdev.at/products/vusb/index.html
[14:47:54] <twnqx> yeah, but this is just the software core :P
[14:48:45] <twnqx> as in, you can implement yourself if you want more than just i2c - the i2c one is just an example using it
[14:49:23] <KotH> rotfl
[14:49:27] <KotH> that device is cool
[14:49:35] <jai> on an related note, why isn't the h.264 timestamps related gsoc task not mentioned on the wiki?
[14:49:55] * KotH mentions that for proper operation, you have to be able to connect/disconnect the pull up resistor on DM/DP
[14:50:34] <KotH> jai: maybe it's waiting for you to write an entry for it? :)
[14:51:11] <twnqx> KotH: this is an example of "works mostly" :P
[14:52:27] <jai> KotH: meh :)
[14:52:32] <DonDiego> maybe you guys wish to move this to #wherever-this-chatter-is-ontopic
[14:52:33] <DonDiego> :)
[14:53:46] <thresh> there will be no purpose for having this channel then :p
[14:55:01] <KotH> DonDiego: do you mean #pathscale? ;)
[14:55:22] <jai> thresh: btw, has there been any discussion on the dvd audio support task you guys have on your wiki?
[14:55:35] <jai> thresh: is there any reason that cant be done in fflibs/
[14:55:41] <thresh> personally i have no idea
[14:55:43] <thresh> sorry
[14:55:48] <jai> thresh: k
[15:02:29] <kshishkov> jai: there is - they are now aware about lavf or lavd
[15:02:36] <kshishkov> *not
[15:14:58] <jai> kshishkov: lol
[15:42:54] <kshishkov> jai: please translate http://lleo.aha.ru/na/en/ and I'll try to add it there so we have something to refer those potential GSoCers to
[15:43:35] <jai> kshishkov: LMAO
[15:43:38] <jai> kshishkov++
[15:43:45] <kshishkov> I'm deadly serious
[15:47:47] <nfl> pl correct me if im wrong - if pixfmt is pal8 in a decoder, the palette is loaded in frame.data[1] and frame.data[0] is filled with indices to the palette for each pixel, right?
[15:48:30] <kshishkov> correct
[15:49:02] <nfl> are there any other nitty gritties I need to be aware of?
[15:49:31] <kshishkov> palette format
[15:49:53] <kshishkov> 256 32-bit entries of (R<<24) | (G << 16) | B
[15:49:59] <kshishkov> IIRC
[15:50:16] <kshishkov> it's usually stored in different form in files
[15:50:51] <nfl> thanks
[15:51:02] <kshishkov> np
[15:51:57] <jai> i have to say BBB is a _very_ patient man :)
[15:52:09] <BBB> huh?
[15:52:16] <BBB> oh, regarding that guy?
[15:52:26] <jai> :)
[15:52:42] <BBB> well look there's a 1% chance that he's a good coder with no social skills whatsoever
[15:52:54] <BBB> so maybe let's give him a chance?
[15:53:15] <BBB> I mean, look how quickly he udnerstood the suggestion of sending a new email
[15:53:18] <BBB> it only took 4 tries
[15:53:20] <jai> also a very important skill is being able to scavenge information/rtfs etc :)
[15:53:22] <mru> lol
[15:53:47] <BBB> imagine how quickly he would pick up the codebase of ffmpeg
[15:54:20] <jai> heh
[15:55:09] <BBB> spyfeng: pathc is ok, I'll leave it for 2-3 days for others to review
[15:55:12] <BBB> spyfeng: then I'll apply
[15:55:19] <BBB> spyfeng: consider yourself ok for the SoC, basically
[15:58:18] <j0sh> there was some chatter here recently about unix domain sockets
[15:58:32] <j0sh> would patches adding in support for that be welcome?
[15:59:07] <BBB> ?
[15:59:39] <j0sh> for IPC, rather than using pipes (impossible to have multiple readers on the same pipe, whereas you can with sockets)
[16:00:19] <j0sh> irc log from feb 8th shows you guys discussed it briefly :)
[16:00:20] <j0sh> viva google
[16:01:25] <KotH> BBB: correct taht 1% to 0.1%... he is from india after all
[16:01:45] <KotH> jai: no offence intended at your side
[16:02:35] <jai> KotH: none taken :)
[16:02:36] <kshishkov> jai is kinda an exception. I think FFmpeg has more patches from Ukrainian coders
[16:02:50] <jai> s/coders/coder
[16:02:57] <jai> unless there are more i'm missing?
[16:04:54] <KotH> jai: btw: are there any good books that explain the culture of india? especialy how to deal with people from there?
[16:05:13] <kshishkov> I can name that Sergiy who did some work on network protocols and Maksym Veremeenko who sent misc patches for minor bugs
[16:05:37] <kshishkov> KotH: yes, written in Britain in Victorian times
[16:05:50] <jai> KotH: perhaps try "In Spite of the Gods: The Strange Rise of Modern India"
[16:06:13] <jai> the author is ed luce
[16:07:00] <kshishkov> and something by Kipling
[16:09:01] <KotH> jai: hmm.. looks like an interesting book... thanks!
[16:11:34] * KotH wonders how much internet-english people from india do understand
[16:11:58] <jai> KotH: no problem, i'm pretty sure you'll find some of the stories/incidents interesting :)
[16:12:33] <jai> i assume internet-english != memes :)
[16:12:55] <kshishkov> no, it's "rtfm lmd!" too
[16:13:28] <Dark_Shikari> they better know our memes too
[16:14:02] <KotH> jai: internet-english = semi-technical simplified english used on the internet
[16:14:32] <KotH> 'cause, i have the feeling that our friend does not understand what "qualification task" means
[16:14:35] <kshishkov> "Hi, my name is Kumar, I play Touhou games. Can I have some task for GSoC 2010 in VideoLAN?"
[16:15:03] <DonDiego> Yuvi: you around?
[16:15:20] <jai> KotH: ah, well a lot of people are pretty well versed with netiquette and english usage, but there are so many of us that it doesnt amount to much :|
[16:15:52] <Dark_Shikari> kshishkov: lol
[16:16:03] <jai> touhou isn't that popular here btw :)
[16:16:04] <Dark_Shikari> you mean "my name is Yakamura"
[16:16:10] <jai> lol
[16:16:13] <DonDiego> are these messages arriving on ffmpeg-devel?
[16:16:19] <DonDiego> they should go to ffmpeg-soc
[16:16:40] <jai> that isnt mentioned anywhere afaik
[16:16:59] <jai> also, more visibility on -devel
[16:17:05] <DonDiego> i just looked it up on the wiki
[16:17:09] <DonDiego> it says -devel
[16:17:22] <jai> exactly
[16:17:31] <DonDiego> jai: more visibility is not good in this case, -devel is too high traffic
[16:17:42] <kshishkov> jai: yes, in a decade there will be more of you than in China
[16:17:45] <DonDiego> there should be a filter installed in front of michael
[16:18:10] <DonDiego> jai: don't you have a problem with the boy/girl ratio?
[16:18:10] <jai> DonDiego: well, i was thinking more in terms that others on the list might have opinions as well...
[16:18:29] <jai> DonDiego: in gsoc??
[16:18:38] <DonDiego> no, in india :)
[16:18:49] <DonDiego> haha, yes, i'm speaking in riddles
[16:18:56] <Dark_Shikari> I have a problem with the indian/everyone else ratio
[16:19:03] <Dark_Shikari> then again, I just reject every indian, so that's rarely an issue
[16:19:10] <DonDiego> BBB: i'll redirect people to ffmpeg-soc on the wiki if you don't mind
[16:19:12] <jai> ah, yeah, in some parts
[16:19:20] <kshishkov> Dark_Shikari: ask for help in CHina
[16:19:20] * DonDiego looks around for other soc mentors
[16:19:21] <BBB> no problem
[16:19:30] <Dark_Shikari> kshishkov: I have a chinese friend that might be applying
[16:19:38] <DonDiego> BBB: also, what about the "new regtest system" task?
[16:19:42] <Dark_Shikari> though he's in the US, so it doesn't count
[16:19:47] <BBB> dondiego: add it
[16:19:50] <BBB> who will mentor it?
[16:19:51] <kshishkov> Dark_Shikari: so what? One even got through already
[16:20:03] <DonDiego> i talked about it with mru, he tells me there is no need
[16:20:07] <jai> if someone is editing the wiki, maybe they might want to add the h,264 timestamps related thing michael said he'd mentor
[16:20:15] <jai> *h.264
[16:20:18] <DonDiego> jai: add it :)
[16:20:21] <Dark_Shikari> kshishkov: so what what
[16:20:33] <jai> DonDiego: don't have the wiki credentials handy :|
[16:20:54] <DonDiego> then add it the next time you have them handy
[16:21:01] <kshishkov> Dark_Shikari: even a couple of good Chinese coders don't overcome the fact there are >1.2 milliard of them
[16:21:01] <jai> k
[16:21:26] <Dark_Shikari> kshishkov: it doesn't matter how many of them if a) they suck or b) they are not here ;)
[16:21:48] <Dark_Shikari> you can have a trillion grains of sand on the beach, but if there's none in the shop, you can't make glass
[16:22:15] <j0sh> for whoever's adding the h264 for gsoc in the wiki, here are some related small tasks: http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2010-February/083177.html
[16:23:16] <BBB> koth: the mentor (baptiste) should approve his qualification task for that particular soc project
[16:23:29] <BBB> koth: we'll do it a bit different than last few years
[16:24:08] <BBB> kshishkov: american word for milliard is billion
[16:24:27] <KotH> BBB: oh.. didnt know
[16:24:29] <KotH> BBB: thanks
[16:24:40] <kshishkov> BBB: why do you shrink it thousandfold?
[16:24:53] <BBB> kshishkov: billion in american isn't billion in european
[16:24:54] <Dark_Shikari> kshishkov: you mean multiply it a thosuandfold
[16:25:00] <Dark_Shikari> in stupid countries, billion == 10^12
[16:25:03] <BBB> europeans made up this weird billion = 1000*milliard
[16:25:07] <Dark_Shikari> in cool countries, billion = 10^9
[16:25:11] <BBB> exactly
[16:25:23] <BBB> europeans also made up this weird word milliard
[16:25:25] <mru> anyone for a game of billiards?
[16:25:27] <jai> j0sh: of course, that could be linked to on the small tasks page as well
[16:25:34] <BBB> lol :)
[16:25:35] <Dark_Shikari> mru: lol
[16:25:36] <KotH> mru: sure..
[16:25:41] <KotH> mru: do you come to zh?
[16:25:41] <mru> strictly speaking, europeans were first
[16:25:52] <kshishkov> BBB: million too, it's from Italian "mille" = "thousand"
[16:25:55] <mru> the yanks couldn't cope with it, so they simplified the language
[16:25:59] <mru> or "lingo" as they say
[16:26:04] <Dark_Shikari> mru: it's called "using metric"
[16:26:05] <Dark_Shikari> ironically
[16:26:08] <Dark_Shikari> really really ironically
[16:26:14] <KotH> rotfl
[16:26:25] <Dark_Shikari> I mean, who the hell thought "milliard" was a good idea
[16:26:33] <mru> germans
[16:26:36] <Dark_Shikari> probably the same guy who likes gallons per mile-fortnight
[16:27:02] <mru> says the guy from the coutry where acre-inches is a unit of volume...
[16:27:06] <KotH> Dark_Shikari: you mean they should have left that *ard away completely?
[16:27:18] <KotH> Dark_Shikari: or do you mean just to leave out the milliard?
[16:27:34] <kshishkov> hmm, "milliard" comes from French
[16:27:50] <kshishkov> so it's stupid word for Americans
[16:28:01] <mru> and then we have the chinese "man" for 10k
[16:28:04] <Dark_Shikari> mru: yes, I'm saying it was the same guy
[16:28:14] <Dark_Shikari> responsible for stupid shit like miles, acres, and milliards
[16:28:18] <kshishkov> mru: greeks had "miria"
[16:28:31] <mru> Dark_Shikari: the US is the only remaining country to use those silly units
[16:28:38] <mru> britain is metric
[16:28:43] <Dark_Shikari> mru: no, a few more do
[16:28:45] <Dark_Shikari> shitty ones
[16:28:45] <mru> both officially and in practice
[16:28:48] <KotH> mru: even in daily life?
[16:28:50] <Dark_Shikari> and yes britain uses imperial
[16:28:54] <Dark_Shikari> actually britain is worse
[16:28:56] <mru> KotH: to a large extent, yes
[16:29:01] <KotH> cool
[16:29:02] <mru> beer is still in pints
[16:29:06] <Dark_Shikari> in america, I'm 175 pounds
[16:29:06] <KotH> didnt take them too long
[16:29:08] <mru> and roads are in miles
[16:29:09] <Dark_Shikari> in britain, I'm 12.5 STONE
[16:29:10] <mru> all else is metric
[16:29:12] <Dark_Shikari> what the fuck is a stone
[16:29:18] <mru> 14 pounds
[16:29:23] <Dark_Shikari> yes, but the point...
[16:29:28] <kshishkov> and guinea is 21 shillings :)
[16:29:30] <Dark_Shikari> lol
[16:29:39] <mru> and how long is a chain, anyone?
[16:29:46] <mru> it's 100 links fwiw
[16:29:48] <kshishkov> all its links
[16:30:05] <KotH> mru: dunno, but it'll break at its weakest link ;)
[16:30:17] <mru> and who knows the difference between a troy ounce and an apothecarie's ounce?
[16:30:27] <KotH> (yes, i'm bored... i dont want to work)
[16:30:29] <mru> no, neither of them is the fluid ounce
[16:30:36] <Dark_Shikari> lol
[16:30:42] <Dark_Shikari> fluid ounce is a measure of volume
[16:30:46] <kshishkov> isn't that suspicious that the only survived non-metric Russian measures are for volume of alcoholic drinks?
[16:30:55] <KotH> kshishkov: same in .jp
[16:30:57] <mru> and don't forget the dram
[16:31:01] <mru> or the grain
[16:31:04] <KotH> kshishkov: sake is still measured in rice cups
[16:31:06] <kshishkov> farlong
[16:31:10] <mru> furlong
[16:31:15] <mru> per fortnight
[16:31:15] <kshishkov> KotH: six-tatami room :P
[16:31:19] <Dark_Shikari> no no no
[16:31:23] <Dark_Shikari> attoparsecs per microfortnight
[16:31:27] <Dark_Shikari> best unit of measure ever.
[16:31:34] <mru> no, the best one is the beard-second
[16:31:37] <KotH> kshishkov: only for japanese style rooms... all others are measured in m^2
[16:31:38] <Dark_Shikari> lol
[16:32:03] <KotH> kshishkov: looks very wierd on house plans when half of the rooms are measured in tatami, and the other half in m^2
[16:32:33] <Dark_Shikari> tatami-stone per microfortnight
[16:33:03] <mru> Dark_Shikari: micro is way too metric
[16:33:46] <kshishkov> mru: and in my homeland they often use hectograms - "tre hecto k\:ott"
[16:35:15] <mru> sweden is the only place I know to use hectograms and decimetres
[16:36:37] <kshishkov> could be
[16:38:23] * mru includes norway in sweden
[16:38:35] <kshishkov> too late for that
[16:38:49] <kshishkov> for 105 years to be precise
[16:39:32] <kshishkov> and 200 years for \:Osterland and part of Norrland (aka Suomi)
[16:40:29] <KotH> mru: we dont use hectogram in .ch, but decimeter is quite comon
[16:40:39] <KotH> mru: same for deziliter
[16:40:56] <kshishkov> decimeter/deciliter is common even here
[16:41:19] <mru> and 2.5 dl ~= 1 cup
[16:41:40] <mru> not to be confused with the old swedish measure 1 kaffekopp == 1.5 dl
[16:42:56] <kshishkov> I though 1 kaffekopp = fem minuter
[16:42:59] <kshishkov> *thought
[16:43:13] <kshishkov> and BTW, how much grams in one pound
[16:43:24] <mru> 454
[16:43:25] <BBB> pound =~ 0.5 kilo
[16:43:26] <mru> ~
[16:43:32] <BBB> ounce ~= 0.1 kilo
[16:45:29] <kierank> nobody's mentioned knots and nm yet
[16:45:38] <kshishkov> knot is mph
[16:45:53] <kshishkov> nautical mile though
[16:45:54] <kierank> you mean nmph
[16:46:04] <kshishkov> 1852metres
[16:46:10] <kshishkov> or something
[16:46:19] <mru> 1 minute of arc at equator iirc
[16:46:43] <mru> so if you're doing one minute per minute...
[16:46:51] <mru> your mind forms a knot
[16:46:53] <mru> hence the name
[16:47:15] <kshishkov> s/forms/spins into/
[16:47:22] <kshishkov> it's angular minute after all
[16:48:31] <KotH> a gordic knot?
[16:49:19] <kshishkov> gordian
[16:49:41] <kshishkov> you should have had major in Classics
[16:50:30] <kierank> so you can learn dead languages...
[16:50:43] <kshishkov> like COBOL
[16:50:52] <kshishkov> or ALGOL-60
[16:51:47] <kierank> well cobol can easily be converted to LOLCODE
[16:54:07] <KotH> kshishkov: there were no majors/minors at the eth
[16:54:13] <KotH> kshishkov: the concept just doesnt fit there :)
[16:54:36] <mru> eth sounds like a proper school
[16:54:40] <kshishkov> KotH: and almost all education concepts don't fit in our system
[16:55:55] <DonDiego> mru: what about adding __BSD_VISIBLE to cppflags for freebsd?
[16:56:00] <DonDiego> the fate box needs it
[16:56:31] <mru> why?
[16:56:35] <mru> sounds wrong
[16:57:00] <DonDiego> it could be wrong, but it won't build without
[16:57:08] <DonDiego> we have similar hacks for solaris and netbsd
[16:57:46] <DonDiego> i've given up hope of fixing the bsd platforms
[16:57:55] <DonDiego> the hack is very small and self-contained
[16:58:03] <DonDiego> http://fate.multimedia.cx/index.php?build_record=205517
[16:58:06] <kshishkov> mru: why idea of adding obscure flags to fix compiling on BSD sounds wrong to you?
[16:58:15] <DonDiego> and it's in use on the fate box anyway..
[16:59:00] <mru> DonDiego: that doesn't mean it's the right thing to do
[16:59:26] <DonDiego> it means that it's done in practice
[16:59:44] <DonDiego> i know it's not the right thing(TM)
[16:59:57] <mru> maybe there's a better solution
[17:00:03] <mru> what breaks without it?
[17:01:27] <DonDiego> the build :)
[17:01:34] <DonDiego> let me fish out the details from a mail..
[17:01:37] <barque> anyone know how to do timing with lbavcodec?
[17:01:41] <barque> libavcodec*
[17:02:01] <barque> like I decode the frames, just wanna know the second/millisecond position it should be played
[17:02:17] <barque> cause decoding is quite fast :P
[17:02:32] <DonDiego> mru: bktr.c, ossaudio and udp stuff
[17:02:39] <mru> not enough details
[17:02:51] <mru> pile all the hacks you want _inside_ bktr.c
[17:02:55] <mru> that's bsd-specific anyway
[17:03:20] <mru> I'm also sorely tempted to simply drop all manner of bsd support
[17:03:28] <mru> they're a huge pita
[17:03:32] <DonDiego> we use global hacks for solaris and netbsd already
[17:03:32] <mru> and nobody uses it anyway
[17:03:39] <mru> and that's wrong
[17:04:13] <DonDiego> i know i'm going back and forth but in the end this dogmatic approach buys us nothing i'm afraid
[17:04:23] <DonDiego> one line in configure is well-contained enough
[17:04:24] <mru> it buys us clean code
[17:04:37] <mru> it can break other things
[17:05:08] <mru> sometimes requesting BSD stuff gives you non-posix variants of some functions
[17:05:16] <DonDiego> it apparently does not in practice
[17:05:22] <mru> you don't know that for sure
[17:05:37] <DonDiego> the fate box does it
[17:05:37] <mru> our test coverage is pretty poor
[17:05:46] <DonDiego> it's necessary for compilation
[17:05:55] <mru> that's not what I'm talking about
[17:06:12] <mru> I'm talking about random functions having their arguments reversed and shit like that
[17:06:16] <barque> oh wait... av_gettime () gives you time in video?
[17:07:05] <mru> show me exactly what breaks without that thing defined
[17:07:23] <DonDiego> i just asked michael kostylev for a log..
[17:07:58] <DonDiego> but i think having hacks on the fate boxes is pointless
[17:08:06] <mru> yes
[17:08:14] <DonDiego> if they are used on the fate boxes they might as well be in configure
[17:08:15] <mru> but that doesn't justify dirty hacks in the source
[17:08:34] <DonDiego> it's just one line in configure
[17:08:44] <mru> that impacts every last line of C code
[17:08:46] <mru> no, thanks
[17:09:24] <DonDiego> then drop those fate boxes?
[17:09:52] <mru> I spent several hours digging once before discovering that some function had the arguments reversed between tru64 and solaris
[17:10:11] <mru> it still compiled
[17:10:15] <mru> not even a warning
[17:10:47] <mru> tru64 is a bit bsd-ish
[17:11:44] <DonDiego> i know that this is all true
[17:11:52] <mru> then stop arguing
[17:11:57] <DonDiego> but right now we have a split personality
[17:12:08] <DonDiego> these hacks are used on the fate boxes
[17:12:13] <mru> just get me those error logs
[17:12:29] <DonDiego> i'll pass them along once they arrive
[17:27:08] <mru> the things you find in the referrer stats...
[17:27:11] <mru> http://en.wikipedia.org/w/index.php?title=Bit_field&action=historysubmit&diff=351969241&oldid=351753763
[17:39:11] <DonDiego> Yuvi: you around?
[17:46:38] <kshishkov> hmm, s/popular/less well optimized/ was a good edit
[18:17:41] <ramiro> mru: lots of cheappills.com in referrer stats too?
[18:18:06] <mru> quite a bit of spam, yes
[18:18:22] <kshishkov> spam spam spam eggs sausages and spam
[18:18:49] <mru> nowadays it's spam spam spam pills watches and spam
[18:19:34] <BBB> merbanan's blog is full of pill-ads
[18:19:38] <BBB> unfortunately :(
[18:19:49] <mru> comment spam?
[18:19:50] <kshishkov> he should post something about ATRAC there
[18:20:11] <BBB> yes
[18:20:34] <mru> no spam blocker?
[18:21:03] <kshishkov> he neglects his blog
[18:21:35] <DonDiego> \o/
[18:21:51] <DonDiego> libsdl 1.3 preview does not make ffplay crash immediately on os x
[18:22:00] <BBB> ?
[18:22:05] <BBB> ffplay works for me
[18:22:06] <DonDiego> i can quickly test samples again, yay
[18:22:12] <DonDiego> you are not on ppc
[18:22:15] <BBB> oh
[18:22:16] <BBB> right
[18:22:18] <kshishkov> FFplay works for me too
[18:22:22] <BBB> I have to admit that the visualizer is ... weird
[18:22:29] <BBB> I've always wondered if that's a feature or a bug
[18:22:31] <DonDiego> which libsdl version?
[18:22:37] <DonDiego> i use macports
[18:22:39] <BBB> mike melanson's blog showed the same img as mine though
[18:22:42] <kshishkov> don't remember
[18:23:14] <kshishkov> BBB: you've written something about GSoC2010 then
[18:23:26] <kshishkov> what about the picture in my last blog post?
[18:24:30] <BBB> http://multimedia.cx/eggs/wp-content/uploads/2010/02/ffplay-spectrum-wma-voice.jpg
[18:24:32] <BBB> that one
[18:24:34] <BBB> weird
[18:28:04] <ramiro> BBB: I get the same thing from ffplay
[18:28:55] <kshishkov> at least it allows you to recognize instruments from voice without even sound turned on ;)
[18:31:05] <BBB> it looks so weird
[18:31:29] <kshishkov> only for the first time
[18:31:38] <pJok> kshishkov, sounds like you really have a nice transport system in ukraine ;)
[18:32:36] <ramiro> hmm, we are talking about the weird green lines, right?
[18:33:23] <kshishkov> pJok: of course. Best Czechoslovakian transport from 70s and some Soviet and Chinese crap
[18:33:48] <BBB> ramiro: the red also confuses me
[18:33:54] <justlooking> wbs, bbb do you intend patching ffmpeg to parse generic youtube URLs to get the real rtsp2.youtube.com links such as http://www.youtube.com/watch?v=anwy2MPT5RE , id ask you question on #ffmpeg but you not one that channel.
[18:34:24] <kshishkov> green - both channels are the same there, red/white - left/right channel only
[18:34:32] <kshishkov> it's quite intuitive
[18:34:36] <BBB> justlooking: that's almost playlist parsing...
[18:34:42] <BBB> justlooking: so I'm not sure
[18:34:51] <BBB> justlooking: but it should be easy to do and I could look at the patch
[18:34:57] <BBB> I might conclude it doesn't belong in ffmpeg though
[18:35:22] <jai> write a script and pass the url to ffmpeg instead
[18:36:48] <kshishkov> BBB: you actually have so-called inner ear which acts as a frequency analyzer. Use it compare FFplay output.
[18:39:24] <jai> O_O
[18:39:56] <BBB> my inner ear is composed of tissue, it appears after a quick rough dissection
[18:40:05] <BBB> doesn't look at all like that window in ffplay
[18:40:18] <kshishkov> learn physiology
[18:41:10] <kshishkov> and if you can't compare sound frequencies to FFplay output you need your neuron connections patched
[18:43:04] <BBB> so what is the green line again?
[18:43:21] <kshishkov> where frequencies for both channels are the same
[18:43:27] * justlooking thought the " frequency analyzer" was the gray matter and the inner ear was the analogue sensor.
[18:44:39] <pJok> kshishkov, isn't that when you say patch welcome?
[18:45:12] * pJok wonders how his hearing will be after april 9th
[18:45:14] <kshishkov> pJok: yes but in what format?
[18:45:28] <pJok> kshishkov, something like a fpga probably
[18:46:37] <kshishkov> well, there's only one man here who is supposed to know about human organism
[18:46:42] * kshishkov looks at BBB
[18:47:03] <_av500_> big buck bunny?
[18:48:11] <BBB> I know little about the ear
[18:48:13] <BBB> more about brain
[18:49:28] <kshishkov> still you should know more than me
[18:51:08] <BBB> unlikely
[18:51:16] <BBB> zero divided by zero is undefined
[18:51:36] <kshishkov> cmp, not div
[18:59:14] <justlooking> BTW , FFmpeg version SVN-r22644 and > from http://ffmpeg.arrozcru.org/autobuilds/ lock up after reporting drop=12 on ffmpeg -i rtsp://rtsp2.youtube.com/CjcLENy73wIaLgkCo6RfCuxKFhMYDSANFEIQdGVzdC1kb21pbml0eS0wMUgGUglwbGF5bGlzdHMM/0/0/0/video.3gp test.mp4, i assume its not a bad file!
[19:00:00] <DonDiego> BBB: what about the regtest soc task? you added it in the first place..
[19:00:08] <BBB> ?
[19:00:12] <BBB> when?
[19:00:43] <BBB> I didn't add that one
[19:00:57] <BBB> I copied it from the 2009 page
[19:12:15] <DonDiego> ok, i will just remove it since it's just cruft apparently
[19:12:30] <DonDiego> unless somebody can come up with a good reason to keep it..
[19:16:46] <BBB> DonDiego: feel free to remove
[19:18:50] <bcoudu> hi guys
[19:19:15] <kshishkov> heloo script from truncated nick
[19:19:44] <bcoudurier> better :>
[19:20:58] <kshishkov> yes, probably
[19:21:52] <DonDiego> bcoudurier: hi
[19:26:11] <CIA-24> ffmpeg: bcoudurier * r22670 /trunk/libavcodec/ (avcodec.h libx264.c options.c): Enable more libx264 options, aq, psy, rc lookahead and ssim
[19:39:58] <CIA-24> ffmpeg: bcoudurier * r22671 /trunk/ (29 files in 2 dirs): Enable mbtree by default, and remove it from presets
[19:48:16] <CIA-24> ffmpeg: mstorsjo * r22672 /trunk/libavformat/rtsp.c: Reindent
[19:59:33] <BBB> j0sh: did I just get an empty email from you?
[19:59:41] <BBB> maybe just random spam
[19:59:42] <BBB> ...
[20:08:21] <hyc> so, about rtmp again... can someone list the actual items that need to change in the patch I posted, to make it "acceptable" ?
[20:12:51] <BBB> hyc: link ffmpeg to ssl and integrate it into our rtmp.c code
[20:12:54] <BBB> deal
[20:14:03] <hyc> BBB: that doesn't sound to me like keeping both code paths in parallel
[20:14:33] <BBB> michael is ok with optionally linking ffmpeg to libssl
[20:14:45] <BBB> there's no reason to not integrate rtmpe into ffmpeg
[20:15:44] <hyc> right, but it's a hassle to backport librtmp into ffmpeg's rtmp code. it will take a lot more testing, and doubles the amount of maintenance I have to deal with
[20:15:59] <BBB> right
[20:16:01] <hyc> whereas just dropping librtmp in requires very little additional testing
[20:16:06] <BBB> so you'll drop librtmpe and focus on ffmpeg alone, right?
[20:16:13] <hyc> nope, can't do that
[20:16:20] <hyc> since there are already other apps using librtmp
[20:16:29] <BBB> does any of them not use ffmpeg?
[20:17:02] <hyc> yes, some of them are just streaming tools, like rtmpdump, with no A/V functionality
[20:17:54] <BBB> define "some"
[20:17:57] <BBB> you're giving one
[20:18:22] <hyc> anyway, I was trying to give you guys something, and now you're telling me that in order for it to be "good enough" to accept I have to rewrite the whole thing.
[20:18:48] <hyc> That's pretty strong disincentive for contributions.
[20:19:43] <hyc> re: some - rtmpgw, rtmpsuck, rtmpdump-yle to name a few more
[20:20:25] <wbs> hyc: even if I prefer having everything within the same code base, I do see the situation that librtmp will be better maintained and up to date, so a patch that allows the usage of librtmp optionally instead of the built-in code would be ok for me
[20:21:56] <hyc> wbs: ok. that sounds to me like an autoconf switch and not much more. is that what your're talking about?
[20:22:23] <wbs> hyc: we don't use autoconf, but yes, have a look at how the integration with other external libraries work, and see if you can do something similar
[20:23:06] <hyc> wbs: ok, will look at that
[20:25:45] <wbs> hyc: I hope you understand the general reason for things, though - we prefer stuff written within the existing libav* frameworks, and simply dropping in external code within that doesn't really fit, so if one wants code merged into the project properly, it usually requires quite a bit of rewriting and/or polishing up, to use common routines and designs etc
[20:26:39] <wbs> although all of us would prefer the libavformat rtmp code to be the best one, I don't think anybody can force you to work on that one if you don't want to :-)
[20:27:21] <hyc> wbs: I would have assumed that one of the advantages of the modulear API is that you can drop in external code easily
[20:27:33] <BBB> gstreamer is very modular
[20:27:59] <wbs> hyc: modular yes, but not in that sense. not every single audio/video codec should bring their own half-assed fft for example
[20:28:12] <_av500_> ffmonolithic
[20:28:43] <hyc> wbs: I totally agree with that. if you've got a nicely optimized math lib it'd be stupid not to make the most use of it
[20:35:40] <hyc> and while libavutil may have a couple things we could use, it still isn't enough to obviate linking an entire crypto library in
[20:36:18] <wbs> of course not, and I'm actually a bit doubtful of reimplementing a whole ssl lib, therefore accepting using libssl or similar is a good decision
[20:36:31] <hyc> ok
[20:36:55] <BBB> michael is ok with an optional dep on libssl
[20:37:03] <hyc> I get it, I know where you're coming from
[20:38:08] <hyc> heh, I never read this configure script before, just assumed it was autoconf
[20:38:26] <wbs> it's like a million times better than autoconf
[20:38:38] <hyc> yeah, I can tell ;)
[20:39:44] <hyc> ok, going to add a --enable-librtmp
[20:40:15] <wbs> can't guarantee that others will accept it, but that'd be the normal approach for such a thing at least
[20:40:36] <wbs> and given the current situation, i'd really consider using it
[20:40:53] <hyc> and I guess thereought to be a separate --enable-libssl switch
[20:41:09] <hyc> librtmp would require it, but you could optionally support https then
[20:42:11] <Vitor1001> bcoudurier: was it your commit that broke regtest-ffm?
[20:43:01] <bcoudurier> roh
[20:43:30] <BBB> maybe I broke it, I committed something to ffmdec
[20:44:07] <bcoudurier> well I feel it's the default flags
[20:44:14] <Vitor1001> looking at FATE, it looks like 22699 is ok, 22700 is not
[20:44:14] <bcoudurier> only lavf right ?
[20:44:40] <Vitor1001> right.
[20:46:21] <CIA-24> ffmpeg: bcoudurier * r22673 /trunk/tests/ref/lavf/ffm: 10l, update ref value for ffm since default flags changed and are stored in the file
[21:02:23] <CIA-24> ffmpeg: stefano * r22674 /trunk/libavcodec/wmaprodec.c:
[21:02:23] <CIA-24> ffmpeg: Make wmaprodec.c:decode_init() return AVERROR_INVALIDDATA /
[21:02:23] <CIA-24> ffmpeg: AVERROR_PATCHWELCOME in case of invalid / unsupported number of
[21:02:23] <CIA-24> ffmpeg: channels specified, rather than return AVERROR_NOTSUPP.
[21:09:41] <wbs> ok, popular vote, which formatting style do people prefer for the ff_data_to_hex patch that I sent earlier?
[21:10:07] <wbs> lu_zero preferred the original style, michael was ok with the new one. to apply or not to apply, that's the question
[21:10:25] <wbs> (aka serious bikeshed)
[21:11:11] * pJok sheds some bikes
[21:12:14] <BBB> wbs: let's keep it, I'd like to know why [16] =""; changed the binary
[21:12:16] <BBB> it shouldn't
[21:12:30] <BBB> in particular because the new binary was likely bigger
[21:12:34] <BBB> (was it?)
[21:12:44] <wbs> don't remember, I can check
[21:12:51] <BBB> nah
[21:12:52] <BBB> don't worry
[21:12:55] <wbs> ok
[21:13:04] <BBB> so your curernt patch, is there any case where uri is not rt->control_uri?
[21:13:28] <wbs> yes, for SETUP it's rtsp_st->control_url, and for some case of OPTIONS it's *
[21:13:39] <BBB> ok
[21:14:07] <BBB> then secondly, can we get rid of buf1[] in send_message()?
[21:14:13] <BBB> (by using av_strlcatf())
[21:15:09] <wbs> yes, that should be fully possible (as a completely separate matter from this)
[21:15:28] <BBB> yeah
[21:16:07] <DonDiego> wbs: k&r style of course :)
[21:16:10] <DonDiego> bbl
[21:32:39] <wbs> hyc: the includes for external libraries (librtmp/*) should be with <> instead of ""
[21:33:33] <wbs> except for that, it looks quite ok at a first glance
[21:36:06] <wbs> BBB: cmd isn't unused for the announce case - it contains the Content-Type header
[21:36:22] <BBB> but that's a static string
[21:36:29] <BBB> so you can pass it directly into send_message()
[21:36:36] <wbs> ah, yes, realized it after pressing enter right now ;P
[21:36:37] <BBB> no need to copy it into cmd[], right?
[21:36:41] <BBB> ok :)
[21:37:16] <BBB> and then you can commit it all
[21:37:28] <wbs> and for the if (auth_state) thing - the auth methods check for that themselves so it isn't needed here
[21:37:32] <wbs> unless you want it for clarity
[21:37:39] <BBB> I think it's better for clarity
[21:37:43] <wbs> ok
[21:37:50] <BBB> you could then maybe remove it from the auth methods
[21:37:54] <wbs> extra lines aren't all that expensive anyway
[21:38:02] <BBB> but whichever you like, I'd prefer if the function wasn't called
[21:38:02] <wbs> sure
[21:39:00] <BBB> you had several more patches to be reviewed right?
[21:39:23] <wbs> only the old thread regarding handling of EINTR from select
[21:39:35] <BBB> "[FFmpeg-devel] [PATCH] libavformat: Don't return errors if select is interrupted"
[21:40:18] <wbs> yep, that one
[21:45:05] <nfl> bcoudurier: hi what would be a good qual task for the seeking api gsoc task?
[21:46:59] <BBB> "[FFmpeg-devel] [PATCH] Use the actual RTSP peer IP for RTP sessions" was applied, right?
[21:47:12] <CIA-24> ffmpeg: mstorsjo * r22675 /trunk/libavformat/ (rtsp.c rtsp.h rtspenc.c): Add separate method/url parameters to the rtsp_send_cmd functions
[21:47:13] <wbs> yes, that thread is finished for my part
[21:48:28] <CIA-24> ffmpeg: mstorsjo * r22676 /trunk/libavformat/ (rtsp.c Makefile rtsp.h):
[21:48:28] <CIA-24> ffmpeg: Make RTSP use the generic http authentication code
[21:48:28] <CIA-24> ffmpeg: Still hardcoded to use Basic auth, without parsing the reply headers
[21:49:15] <BBB> yay @ rtspauth
[21:49:46] <CIA-24> ffmpeg: mstorsjo * r22677 /trunk/ (libavformat/rtsp.c ffserver.c libavformat/rtsp.h): Actually parse the auth headers in RTSP
[21:50:33] <CIA-24> ffmpeg: mstorsjo * r22678 /trunk/libavformat/rtsp.c: Don't force basic auth in RTSP, but retry with the server-specified method on failure
[21:51:15] <wbs> yay @ getting patches ok'd :-)
[21:54:33] <CIA-24> ffmpeg: mstorsjo * r22679 /trunk/libavformat/httpauth.c: Remove a redundant null pointer check
[22:05:29] <CIA-24> ffmpeg: mstorsjo * r22680 /trunk/libavformat/rtsp.c: Simplify ff_rtsp_send_cmd_with_content_async, remove an unnecessary buffer
[22:10:39] <nfl> a quick question: where is pal8 used to build rgb output in the ffmpeg code?
[22:11:00] <BBB> libswscale
[22:16:25] <DonDiego> mru: http://www1.mplayerhq.hu/~diego/freebsd.log
[22:16:48] <DonDiego> that's a log of a failing freebsd build, given to me by michael kostylev
[22:17:30] * mru stabs bsd
[22:20:15] <mru> they're all bugs in the system headers
[22:20:55] <DonDiego> yes
[22:21:26] <mru> bktr.c and oss_audio.c fail because the system headers use typedefs without defining them first
[22:21:40] <mru> udp.c fails on IPPROTO_IPV6 which is quite standard
[22:21:57] <CIA-24> ffmpeg: mstorsjo * r22681 /trunk/libavformat/rtsp.c: Reassemble the RTSP URL before replacing hostname with the numerical IP
[22:22:00] <mru> the spec does not say a thing about special triggers needed to make it visible
[22:22:37] <DonDiego> bktr.c could be fixed through __BSD_VISIBLE
[22:22:43] <mru> has anyone asked bsd devs?
[22:22:46] <mru> as if...
[22:22:50] <DonDiego> dunno
[22:22:56] <DonDiego> where to find them?
[22:23:03] <mru> in the dungeon
[22:23:09] <DonDiego> #freebsd?
[22:29:44] <Kovensky> what would make av_seek_frame segfault?
[22:30:25] <Kovensky> Program received signal SIGSEGV, Segmentation fault.
[22:30:25] <Kovensky> 0x00007ffff7724d68 in FFLAVFAudio (this=0x763420, SourceFile=0x7fffffffe236 "campanella_op.mpg", Track=<value optimized out>, Index=0x70e730) at src/core/lavfaudio.cpp:58
[22:30:28] <Kovensky> 58              if (av_seek_frame(FormatContext, AudioTrack, Frames[0].PTS, AVSEEK_FLAG_BACKWARD) < 0)
[22:30:32] <Kovensky> hurf, I'm probably in the wrong channel
[22:35:18] <BBB> finally I submitted a patch myself again
[22:47:17] <CIA-24> ffmpeg: stefano * r22682 /trunk/libavutil/error.h: Use consistent punctuation rules for the error message descriptions.
[22:47:17] <CIA-24> ffmpeg: stefano * r22683 /trunk/libavutil/error.h: Prefer '///< ...' doxygen inline syntax over '/**< ... */'.
[22:47:18] <CIA-24> ffmpeg: stefano * r22684 /trunk/ (5 files in 2 dirs): Implement av_strerror().
[22:50:10] <hyc> mru: there are already several uses of pkg-config in the configure script
[22:53:37] <CIA-24> ffmpeg: stefano * r22685 /trunk/doc/APIchanges: Add APIchanges entry after av_strerror() addition.
[22:54:04] <mru> hyc: that's no excuse for adding more
[22:54:41] <mru> are there any other than for dirac/schrodinger?
[22:55:11] <hyc> hm, no
[22:55:12] <mru> whose retarded devs made it _impossible_ to use otherwise
[22:55:18] <mru> well, you can guess flags
[22:55:31] <hyc> and of course, you actually generate pkg-config files
[22:55:53] <mru> I'd gladly remove that
[22:56:30] <hyc> anyway, not disagreeing with you. I've only had bad experiences with pkgconfi
[22:56:32] <hyc> g
[22:56:47] <hyc> I was just confused at your message, since it appears that your stuff uses and supports it
[22:57:09] <DonDiego> what msg?
[22:58:32] <hyc> the email saying to ignore pkg-config


More information about the FFmpeg-devel-irc mailing list