michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 6.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
rvalue has quit [Ping timeout: 246 seconds]
rvalue has joined #ffmpeg-devel
<Lynne>
jamrial: why does AV_CHANNEL_ORDER_NATIVE support up to 63 channels rather than 64?
kasper93_ has joined #ffmpeg-devel
kasper93 has quit [Ping timeout: 272 seconds]
<jamrial>
Lynne: i guess it's a doxy mistake
<jamrial>
maybe it was done because AV_CH_LAYOUT_NATIVE from the old api is equal to 1ULL << 63
kasper93_ is now known as kasper93
jarthur has quit [Ping timeout: 260 seconds]
<another|>
heh. read some of the issues on BtbN's build repo again. mildly interesting.
<jamrial>
Lynne: other than the doxy, masked_description() needs to be fixed
thilo has quit [Ping timeout: 255 seconds]
thilo has joined #ffmpeg-devel
lexano has quit [Ping timeout: 268 seconds]
thilo has quit [Ping timeout: 255 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
cone-662 has quit [Quit: transmission timeout]
Kei_N has quit [Read error: Connection reset by peer]
jarthur has quit [Quit: jarthur]
AbleBacon has quit [Read error: Connection reset by peer]
jarthur has joined #ffmpeg-devel
<Lynne>
Daemon404: correction, you could have SIX different ways of preroll signalling, all in the same file
<Lynne>
USAC NIHs another preroll extension flag into its generic extension signalling (each extension must always have a way of coding other extensions into itself)
Kei_N has joined #ffmpeg-devel
jamrial has quit []
<Lynne>
just in case you were wondering, yes, the extensions that the usac extension codes also have a mechanism to pack *other* extensions into them
<mkver>
frankplow: There are a bunch of integer overflows triggered by the VVC conformance fate tests. Is wraparound intended here (i.e. should this be silenced by using unsigned intermediate types) or are these real bugs?
thilo has quit [Quit: WeeChat 2.8]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
Guest38 has joined #ffmpeg-devel
<frankplow>
mkver: Do you have a link handy to a FATE run showing this?
<frankplow>
Yeah I looked at ubitux2, I can’t reproduce locally yet with clang-usan as well
<mkver>
You can't reproduce with clang-usan? I could, last time I tried clang ubsan.
<frankplow>
mkver: Not on a first attempt, I will have another look when I have some more time this evening
HarshK23 has joined #ffmpeg-devel
<mkver>
frankplow: Strangely I can no longer reproduce the errors with clang-ubsan; the vvc-conformance tests failed a week ago with it.
Guest38 has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
Guest38 has joined #ffmpeg-devel
<thardin>
looking at LL-HLS it appears what it really is is just shorter chunks and EXT-X-PRELOAD-HINT
<thardin>
where the latter allows working out the URL of the next chunk, and the server to use HTTP long polling to send the chunk ASAP
<JEEB>
yea
<thardin>
or even HTTP/2 PUSH
<JEEB>
it is all about shorter segments that don't start on a random access point in the comments
<JEEB>
the HTTP/2 stuff I think they partially gave up on due to push-back from CDN vendors
<thardin>
people still struggling with pub/sub
<Sean_McG>
thanks everybody who looked into the UBsan issues yesterday, much appreciated. :)
<Sean_McG>
mkver: I have to head off to work shortly, I'll check the dxt patch on BE later today unless someone gets to it first
<Sean_McG>
also wow, I wonder what is wrong on the valgrind FATE nodes
jamrial has joined #ffmpeg-devel
<Sean_McG>
a few weeks ago people made noise on the mailing list about C17 -- do we still need/want a FATE node for that? I'd be OK with setting one up
<JEEB>
yea I think as RHEL7 finally dies we might be moving there, as I think that's one of the few systems still supported that (by default) doesn't have a C17 capable compiler
<JEEB>
although I'm pretty sure there's a developer repo which contains newer tooling
<JEEB>
since I've seen similar stuff for CentOS
<mkver>
What does "need a FATE node for C17" mean?
<mkver>
Do you mean one where the default -std=c11 is replaced by -std=c17?
<JEEB>
that was my initial expectation with a comment like that, although I was mostly commenting about C17 and not regarding the fate node stuff
cone-355 has quit [Quit: transmission timeout]
<Sean_McG>
mkver: basically, yes
<Sean_McG>
anyways, off for now.. have a good day everyone
<haasn>
has anybody ever attempted unifying vf_showinfo and ffprobe?
<elenril>
not that I know
lexano has joined #ffmpeg-devel
<mkver>
Isn't stuff like that part of NG's plans and part of the reason for his AVWriter?
<elenril>
I don't think there was any actual code in that direction
Guest38 has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
<Lynne>
the fixed-point decoder is somewhat to blame for this, and SBR being tacked on
<Lynne>
and the million and one aac extensions all sharing the same file
<mkver>
How did the fixed-point decoder affect the decoder?
<cone-232>
ffmpeg Timo Rothenpieler master:7945d30e91b9: avutil/mem: limit alignment to maximum simd align
<mkver>
s/decoder\?/encoder\?/
<Lynne>
encoder's float-only, it just affects the decoder by having lots of ifdefs
SystemError has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg-devel
jarthur has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
Krowl has joined #ffmpeg-devel
<Sean_McG>
#ifdef soup is never fun
<Lynne>
Sean_McG: yeah, so, the encoder uses ff_aac_pred_sfb_max, which indicates the maximum possible swb per samplerate
<Lynne>
but the actual maximum number of swbs is limited to ff_aac_pred_sfb_max
<Lynne>
since the encoder reused the decoder's structs, it was stuck with 41
<Lynne>
you should do FFMIN, but copy what the other loops in the same file do: " const int pmax = FFMIN(sce->ics.max_sfb, ff_aac_pred_sfb_max[s->samplerate_index]);"
<Lynne>
and iterate over pmax
System_Error has joined #ffmpeg-devel
<Sean_McG>
can we be sure that "s->samplerate_index <= 41" ?
<Sean_McG>
oh... that's the array with the constraints, sorry... it can be less than 41 as well
<rodeo>
this line is interesting
<rodeo>
if (sfb < PRED_SFB_START || sfb > pmax || sum != 2) {
<Sean_McG>
I wondered about that -- PRED_SFB_START == 10, I believe
<rodeo>
seems to explicitly handle a case where sfb > pmax
<rodeo>
why, I have no idea, and RESTORE_PRED looks like it will use sce0/1->ics.prediction_used with the same out of array bounds sfb
compn has quit [Read error: Connection reset by peer]
Guest38 has joined #ffmpeg-devel
compn has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
iive has joined #ffmpeg-devel
Guest38 has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
Guest38 has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
Guest38 has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
Guest38 has joined #ffmpeg-devel
Guest38 has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
Guest38 has joined #ffmpeg-devel
b50d has joined #ffmpeg-devel
cone-232 has quit [Quit: transmission timeout]
cone-598 has joined #ffmpeg-devel
<cone-598>
ffmpeg Andreas Rheinhardt master:0c204ce9f6b6: avcodec/dxvenc: Use proper alignment, write endian-independent output
<JEEB>
there, posted v6 of the avctx side data patch \o/
<Sean_McG>
\o/
omegatron has quit [Quit: Power is a curious thing. It can be contained, hidden, locked away, and yet it always breaks free.]
Guest38 has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
<cone-598>
ffmpeg Cosmin Stejerean master:69dd1ce610fc: avcodec/libsvtav1: send the EOS signal without a one frame delay to allow for the library to operate in a low-delay mode
<cone-598>
ffmpeg Gyan Doshi release/6.1:25abb63bfc4f: avcodec/libsvtav1: add version guard for external param
<cone-598>
ffmpeg Cosmin Stejerean release/6.1:33efa50fa450: avcodec/libsvtav1: send the EOS signal without a one frame delay to allow for the library to operate in a low-delay mode
wbs has quit [Ping timeout: 260 seconds]
wbs has joined #ffmpeg-devel
Livio has joined #ffmpeg-devel
Gramner has quit [Read error: Connection reset by peer]
System_Error has quit [Remote host closed the connection]
kurosu has quit [Quit: Connection closed for inactivity]