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 7.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
System_Error has quit [Remote host closed the connection]
cone-224 has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
bpmedley has quit [Ping timeout: 244 seconds]
<elenril>
BtbN: still seem to be some missing perl packages on vote server
System_Error has joined #ffmpeg-devel
<elenril>
Can't locate JSON.pm in @INC (you may need to install the JSON module) (@INC contains: /usr/lib/cgi-bin/civs /etc/perl /usr/local/lib/x86_64-linux-gnu/perl/5.34.0 /usr/local/share/perl/5.34.0 /usr/lib/x86_64-linux-gnu/perl5/5.34 /usr/share/perl5 /usr/lib/x86_64-linux-gnu/perl-base /usr/lib/x86_64-linux-gnu/perl/5.34 /usr/share/perl/5.34 /usr/local/lib/site_perl) at
<elenril>
/usr/lib/cgi-bin/civs/votecore.pm line 4.
bpmedley has joined #ffmpeg-devel
<compnn>
if it only has technically one image, is it a 1-dimensional video?
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 265 seconds]
realies has quit [Quit: ~]
realies has joined #ffmpeg-devel
cone-291 has joined #ffmpeg-devel
<cone-291>
ffmpeg Rémi Denis-Courmont master:bd226fdd743c: lavc/h264dsp: R-V V intra loop filter
ngaullier has joined #ffmpeg-devel
Sean_McG has quit [Quit: Lost terminal]
Sean_McG has joined #ffmpeg-devel
mateo` has quit [Quit: WeeChat 4.4.4]
mateo` has joined #ffmpeg-devel
<BtbN>
elenril: installed libjson-perl. That error didn't get logged, so I can't easily check if there's more
cone-291 has quit [Quit: transmission timeout]
<wbs>
elenril: the w32pthreads fix, what do you think about backporting that? I mean, the bug is present in all older versions as well - we don't have tests that trigger it quite as much there, but the bug is still present
ccawley2011 has quit [Read error: Connection reset by peer]
Martchus_ has quit [Ping timeout: 276 seconds]
ccawley2011 has joined #ffmpeg-devel
Martchus has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
^Neo has quit [Changing host]
^Neo has joined #ffmpeg-devel
<haasn>
the documentation for sws_scale_frame() claims that it return 0 on success, but the actual implementation (since it began existing) returns the number of scaled lines
<haasn>
should we fix the documentation or the return value?
<cone-422>
ffmpeg Bin Peng master:decc9e643cc3: lavc/aarch64: Fix ff_pred8x8_plane_neon_10
<cone-422>
ffmpeg Bin Peng master:72a3656e8468: lavc/aarch64: Fix ff_pred16x16_plane_neon_10
<wbs>
haasn: the documentation, IMO. nobody can have relied on the old documented behaviour in this case
<jamrial>
wbs: my guess is that people just checked for < 0 and aborted if so, meaning 0 or > 0 was assumed to be success
<haasn>
my own swscale test framework relied on the documented behavior, which is how I noticed this bug :)
<haasn>
it had a check for !ret to continue
<jamrial>
i'd update the implementation, yes
<wbs>
jamrial: ah, right, that's also possible
<jamrial>
this is why stating "Success if >= 0" is more future proof and extensible :p
<jamrial>
and not just 0
<JEEB>
I think the other swscale function did amount of scaled lines, so probably sws_scale_frame got it from there
rvalue has quit [Read error: Connection reset by peer]
<haasn>
michaelni: ramiro: maybe one of you has an opinion on the many, many open trac issues all revolving around the general theme of "swscale asm routines do out of bound accesses when my buffers are not aligned"
<haasn>
it seems that the only robust fix to these bugs is to entirely disable asm for unaligned buffers
<haasn>
we already print a warning
<haasn>
alternatively we could create an aligned temp buffer and memcpy into the user provided one
<haasn>
this may be faster
<michaelni>
haasn, "Out of bounds" ? you mean read over the last line and last pixel ? use C code for the last line
<michaelni>
or use C code for the rightmost columns of pixels
ccawley2011 has quit [Read error: Connection reset by peer]
ccawley2011 has joined #ffmpeg-devel
ccawley2011_ has joined #ffmpeg-devel
ccawley2011__ has joined #ffmpeg-devel
ccawley2011 has quit [Ping timeout: 265 seconds]
ccawley2011 has joined #ffmpeg-devel
ccawley2011_ has quit [Ping timeout: 265 seconds]
ccawley2011_ has joined #ffmpeg-devel
ccawley2011__ has quit [Ping timeout: 265 seconds]
ccawley2011 has quit [Ping timeout: 265 seconds]
cone-422 has quit [Quit: transmission timeout]
IndecisiveTurtle has joined #ffmpeg-devel
<haasn>
problem in our current design is that we are so many levels of abstraction deep that it's hard to find the right place to retrieve the matching C function
<haasn>
or rather, I would argue that the asm wrappers should already be doing this internally
<haasn>
I'm leaning more and more towards scrapping the existing input/output functions and doing a more principled rewrite
IndecisiveTurtle has quit [Quit: IndecisiveTurtle]
esu has quit [Quit: Esu]
esu has joined #ffmpeg-devel
King_DuckZ has joined #ffmpeg-devel
<Marth64>
congrats new TC
mkver has quit [Ping timeout: 252 seconds]
<beastd>
yeah, grats from me as well at new TC
nasso has quit [Quit: nasso]
lemourin has joined #ffmpeg-devel
lemourin has quit [Killed (platinum.libera.chat (Nickname regained by services))]
King_DuckZ has left #ffmpeg-devel [See you later]
JesseTG has joined #ffmpeg-devel
JesseTG has left #ffmpeg-devel [Leaving]
ngaullier has quit [Remote host closed the connection]
<BtbN>
elenril: I've been going over that patch twice now, and except for one instance of &&/|| at the beginning instead of end of the line, I can't find anything you mentioned
<BtbN>
was the comment to the entire series in general? Or really just patch 2?
<BtbN>
Is there actually some faster way to check massive list of strncmp()? Like, one that iterates over the string that's being searched only once, and compares to all possible matches?