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
jafa has quit [Ping timeout: 272 seconds]
rvalue has quit [Ping timeout: 268 seconds]
kurufu has quit [Remote host closed the connection]
kurufu has joined #ffmpeg-devel
rvalue has joined #ffmpeg-devel
<Sean_McG>
mkver: aye, so I guess it's the AltiVec DSP then, the ppc-linux-gcc-13.2-ubsan-no_altivec-qemu FATE node confirms that
HarshK23 has quit [Quit: Connection closed for inactivity]
<Sean_McG>
one more reason for me to learn AltiVec & VSX
psykose has joined #ffmpeg-devel
<Lynne>
is this about a fate failure on ppc?
<Sean_McG>
aye, avid-meridian
<Lynne>
I did submit a patch for it, not sure if I merged it, I lost access to the machine back then and had to guess
<Lynne>
oh? it's not a failure in swscale?
<Lynne>
that's where the last 2 issues were
<Sean_McG>
no, I don't think so -- the test output does not seem to point there
<Sean_McG>
the test decodes to rawvideo, I wonder if this is just an LE/BE issue
<Lynne>
ah, big endian ppc, swscale only runs on little endian
<Lynne>
*the problematic parts did
<Sean_McG>
oh? I was almost certain it had AltiVec acceleration... I don't think it has an LE variant since they were always BE back then
kurufu has quit [Remote host closed the connection]
<Lynne>
I was careful and only removed the LE variant which had issues IIRC
kurufu has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 260 seconds]
IndecisiveTurtle has quit [Ping timeout: 268 seconds]
jafa has joined #ffmpeg-devel
<Sean_McG>
a part of me wonders if we should keep the BE AltiVec stuff around -- POWER8/9/10 default configurations are LE nowadays (although I understand they all can still do BE -- it's just the "tyranny of the default")
<Sean_McG>
anyways, I'mma go play some video games
System_Error has joined #ffmpeg-devel
deus0ww has quit [Ping timeout: 255 seconds]
deus0ww has joined #ffmpeg-devel
jafa has quit [Remote host closed the connection]
jafa has joined #ffmpeg-devel
thilo has quit [Ping timeout: 256 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
MisterMinister has joined #ffmpeg-devel
Kei_N_ has quit [Read error: Connection reset by peer]
<Lynne>
github disabled the xz repo due to a tos violation
<aaabbb>
and suspended the maintainers
<Lynne>
I don't think someone writing offensive (as an antonym of defensive) code should cause a suspension, even in this case
<Lynne>
especially because the offensive part was in the maintainer-uploaded tar releases, not the repo
<Lynne>
it's their rules, but I should setup a cgit
<BtbN>
Large parts of it were in the repo
<aaabbb>
much of it was in the repo itself
<aaabbb>
yea
<aaabbb>
both were required for the exploit to fully function iirc
<another|>
Lynne: where are you getting the info that maintainer accounts are suspended?
<Lynne>
I've heard nothing about that, aaabbb apparently does
<Lynne>
earlier, I read that it would be a good idea to screenshot every single commit that account did because they may suspend themselves, I never thought github would be the ones doing the suspension
<aaabbb>
all comes from #tukaani
<aaabbb>
Lynne: apparently there is an ongoing government investigation
<another|>
ah, sorry. same colour
<aaabbb>
according to an update on openwal
<Lynne>
openwall? I haven't read anything about a government investigation there
<aaabbb>
that's just what i heard in #tukaani, someone linked to the relevant openwall discussion hours ago but i didn't open it
MisterMinister has quit [Ping timeout: 255 seconds]
Martchus_ has quit [Ping timeout: 246 seconds]
Martchus has joined #ffmpeg-devel
cone-323 has joined #ffmpeg-devel
<cone-323>
ffmpeg Sean McGovern master:f63a87629e52: fate: fix sub2video_{basic, time_limited} on big-endian targets
<cone-323>
ffmpeg Andreas Rheinhardt master:7d9d40069555: postproc/postprocess: Don't generally include arch-specific headers
<cone-323>
ffmpeg Andreas Rheinhardt master:4b44b5eaf0ad: swscale/swscale_internal: Only include altivec header iff HAVE_ALTIVEC
<cone-323>
ffmpeg Andreas Rheinhardt master:ac14d682777a: avcodec/h264_refs: Rewrite code to make control flow clearer
<cone-323>
ffmpeg Andreas Rheinhardt master:dc7a60529c27: avcodec/ratecontrol: Use forward declaration for AVExpr
<cone-323>
ffmpeg Andreas Rheinhardt master:348461e550f8: avcodec/h264_refs: Use smaller scope, don't use av_uninit
<cone-323>
ffmpeg Andreas Rheinhardt master:ba7980d9c0d1: avutil/opt: Avoid av_uninit
rvalue has quit [Ping timeout: 268 seconds]
geoffhill has quit [Quit: geoffhill]
mkver has quit [Ping timeout: 268 seconds]
psykose has quit [Remote host closed the connection]
rvalue has joined #ffmpeg-devel
<thardin>
speaking of parsing and the xz backdoor, some people on lobste.rs are mentioning oil shell, which seems like a saner shell
<thardin>
not sure if it guarantees not being able to sneak in commands via variables though
<Lynne>
I've gotten somewhat used to using eshell in emacs
<Lynne>
how would a different shell help, though? the sneaky part was in the build script, which was m4, wasn't it?
<thardin>
oh, it looked like sh to me
<thardin>
anyway the point is mostly the same
<thardin>
compare system() to execv()
<thardin>
the latter can guarantee this can't happen via suitable typing
cone-323 has quit [Quit: transmission timeout]
<nevcairiel>
finding reasons in shell or m4 parsing seems pointless, when someone has access to the code to put in whatever, they'll find a way to make it look innocent enough
<thardin>
TIL ${####} is valid shell syntax
Mister_D has quit [Ping timeout: 268 seconds]
kurosu has joined #ffmpeg-devel
<elenril>
Actual votes cast thus far: 20
<elenril>
still below 50%
<thardin>
pal: can HTJ2K have ROI?
<thardin>
and should we complain about roi_shift != 0 for HT?
<thardin>
see the latest htj2kdec patches
rajivharlalka has joined #ffmpeg-devel
<Lynne>
no way it doesn't, it should have deltaq per block
<thardin>
hm?
<Lynne>
you need deltaq to do ROI (unless ROI stands for something other than region of interest)
<Gramner>
hmm, maybe ffmpeg shouldn't be linking to liblzma
<Lynne>
my second machine had the bad xz version, I did notice that it was a little bit slower to ssh into, but not significantly
Teukka has quit [Read error: Connection reset by peer]
Teukka has joined #ffmpeg-devel
Teukka has quit [Changing host]
Teukka has joined #ffmpeg-devel
<elenril>
i never notice slow ssh login thanks to dnssec
<Lynne>
because it makes everything super slow?
<elenril>
cold logins tend to be slow because of the validation
jamrial has joined #ffmpeg-devel
<haasn>
Lynne: by about 0.4ms?
<Lynne>
err, dnssec or the xz malicious code?
<ePirat>
elenril, how can I check if I am in the GA (so should have gotten the vote mail) or not?
<elenril>
ePirat: ./tools/general_assembly.pl
<elenril>
and you should have
System_Error has quit [Remote host closed the connection]
SystemError has joined #ffmpeg-devel
<elenril>
did you not?
<ePirat>
elenril, just found it
<ePirat>
it did not match the search query I used for it because it comes from mail.de
<ePirat>
while I expected it to be from ffmpeg.org
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
<kierank>
where is durandal
cone-661 has joined #ffmpeg-devel
<cone-661>
ffmpeg James Almer master:f8fbec8686d4: avutil/frame: use the same data information as the source entry when cloning side data
<cone-661>
ffmpeg James Almer release/7.0:799a7200ee60: avutil/frame: use the same data information as the source entry when cloning side data
<kurosu>
Lynne: kind of remembering Mans trying to build one
<Lynne>
ah, I can see why you'd want to
<Lynne>
even though zlib is everywhere, it isn't a posix-level standard
mkver has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
Livio has joined #ffmpeg-devel
iive has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
lexano has joined #ffmpeg-devel
Marth64 has joined #ffmpeg-devel
Livio has quit [Ping timeout: 268 seconds]
Raz- has joined #ffmpeg-devel
<cone-661>
ffmpeg James Almer master:e37b233ee2c0: Revert "avformat/mov: ignore item boxes for animated heif"
<cone-661>
ffmpeg James Almer release/7.0:5a3b625dbc68: Revert "avformat/mov: ignore item boxes for animated heif"
<haasn>
Lynne: the xz code
<haasn>
Is there any issue to depend *locally* on the sizeof() a struct AVFoo whose size is not part of the ABI? For example, struct MyPrivateCtx { struct AVFoo foo; } init() { myctx->foo = *some_external_foo; } encode_frame() { /* use myctx->foo.fields */ }}
<haasn>
If the field gets extended by newer libavutil, my code will continue working, as I will simply ignore the fields I don't know about; since the struct never "leaves" myctx there is no loss of extra fields
<nevcairiel>
with locally you mean outside of the library? the problem isnt you ignoring the fields, but the library trying to write to them
<haasn>
with locally I mean inside a single encoder in libavcodec/
<haasn>
i.e., the data never leaves the encoder again
MisterMinister has joined #ffmpeg-devel
<BtbN>
Only if the struct lives in the same library
<elenril>
a potential problem is if the struct later develops some fields that it expects to be initialized in its constructor
<elenril>
or, given your example, any allocated fields
<elenril>
I'd avoid it unless the alternative is significantly worse
<haasn>
elenril: but the code using this field could not possibly access any of the new fields
<haasn>
how would it know about them if they did not exist at compile time
<haasn>
this struct*
<haasn>
your concern is only valid if you pass &foo to some other external API call
<haasn>
but I don't, I just want to make a local cache of data that I need internally
<haasn>
and in this case sizeof(AVFoo) < sizeof(void*)
<haasn>
and I would really rather avoid the possibility of an OOM condition on any update
<haasn>
I mean one alternative would be to just duplicate the struct definition inline
<nevcairiel>
that sounds strictly worse, since it can fully diverge
<haasn>
but it can't, it can only be extended
<haasn>
I mean, unless you use memcpy() instead of assigning the same-name fields
kurosu has quit [Quit: Connection closed for inactivity]
deus0ww has quit [Ping timeout: 264 seconds]
deus0ww has joined #ffmpeg-devel
LaserEyess has quit [Quit: fugg]
rodgort has joined #ffmpeg-devel
<Marth64>
13th gen i7-1360P or Ryzen 7 PRO 6850H? what would you choose?
LaserEyess has joined #ffmpeg-devel
LaserEyess has quit [Changing host]
LaserEyess has joined #ffmpeg-devel
<elenril>
haasn: I suppose in this case it's not so bad
<Sean_McG>
my machines are both AMD: a Threadripper 2950X for my workstation and a Ryzen 5 5600X in my gaming rig
<jamrial>
Marth64: do you care about thunderbolt docks? if so, the i7
<Sean_McG>
yeah the Thunderbolt situation is lousy if you depend on it
<Marth64>
no thunderbolt docking needed, all my other machines are AMD <3, I've narrowed down to 2 "Linux-certified" laptops with these CPUs.
<elenril>
no avx512?
<Marth64>
sorry to bring this here. have been battling this for a week now
<Marth64>
i will check if they have it
<Sean_McG>
I have been flip-flopping on switching my workstation to a Raptor Blackbird with a POWER9, but they are not cheap -- the motherboard & CPU combo alone would cost me close to $5K CAD. But yeah, this is severely off-topic for #ffmpeg-devel
<Marth64>
upgrade season haha
<Sean_McG>
I can't convince myself this isn't just an impulse buy, and an expensive one at that
<Marth64>
i'm replacing a macbook m1, i love the thing but i just feel so limited
<Marth64>
80% of my time is spent remoted to my debian machine. so its a sign that mac isn't for me now
<Marth64>
anyways, good day to you all. be well
<Sean_McG>
you too, Marth
kurosu has joined #ffmpeg-devel
Livio has joined #ffmpeg-devel
Marth64 has quit [Remote host closed the connection]
<haasn>
elenril: turns it out it's already being done anyway, by AVDOVIRpuDataHeader header;
rvalue has quit [Ping timeout: 255 seconds]
___nick___ has joined #ffmpeg-devel
rvalue has joined #ffmpeg-devel
AbleBacon has joined #ffmpeg-devel
___nick___ has quit [Ping timeout: 256 seconds]
cone-661 has quit [Quit: transmission timeout]
jafa has left #ffmpeg-devel [Leaving]
Livio has quit [Ping timeout: 255 seconds]
<Lynne>
haasn: no, by about 0.5s
SystemError has quit [Remote host closed the connection]
SystemError has joined #ffmpeg-devel
cone-795 has joined #ffmpeg-devel
<cone-795>
ffmpeg Andreas Rheinhardt master:ff215d955908: fate/image: Fix EXR tests on big endian
<jamrial>
even the old "successful" runs were failures, but for some reason fate shows it as green
SystemError has quit [Remote host closed the connection]
Livio has joined #ffmpeg-devel
<mkver>
Wait, it looks like you already have retired it; the website has just not dropped it yet. Nevermind then.
SystemError has joined #ffmpeg-devel
<mkver>
jamrial: The successful runs were successful; ffmpeg relies on threads nowadays, therefore most test aren't run, but those that are run succeeded.
<jamrial>
mkver: if you look at the logs, it says it failed
psykose has joined #ffmpeg-devel
<michaelni>
mkver, yes its supposed to not run anymore
jnbek has joined #ffmpeg-devel
jnbek has quit [Quit: kthx]
jnbek has joined #ffmpeg-devel
<andrewrk>
is it better to use av_buffersink_get_samples() when the user might want to consume only part of an AVFrame, or for the user to possibly keep a partially consumed AVFrame around themselves?
<cone-795>
ffmpeg Marton Balint master:2df2b4067ed0: avfilter/buffersrc: fix overriding unknown channel layouts with negotiated one
<andrewrk>
looks like AV_BUFFERSINK_FLAG_PEEK addresses this use case
<cone-795>
ffmpeg Marton Balint release/7.0:7ed9ad3467e1: avfilter/buffersrc: fix overriding unknown channel layouts with negotiated one
<andrewrk>
hi michaelni. it's been almost a decade since I was involved in ffmpeg! happy to see you're still around, and that libav project has joined forces
<kierank>
andrewrk: happy to see you found fame and hopefully fortune with Zig
<andrewrk>
hi kierank!
<kierank>
I remember you trying to RE DTS-HD
<andrewrk>
well, if you define "fortune" as "enough money to work on it full time comfortably" then, yes I have found my fortune :)
<andrewrk>
hmm I don't remember doing anything with DTS-HD
<andrewrk>
my ffmpeg-adjacent projects are a (currently bitrotted) music player and a (not very far along) digital audio workstation
<andrewrk>
been rewriting the music player in zig for starters
<andrewrk>
well I also maintain a fork of ffmpeg with the build system ported to zig, which is handy
<kierank>
I thought you tried to reverse engineer a codec at one point
<Sean_McG>
there are far too many build systems
<Sean_McG>
as much as I hate GNU autotools, I generally stick to it as "the devil I know versus the devil I don't"
Raz- has quit [Ping timeout: 256 seconds]
___nick___ has joined #ffmpeg-devel
___nick___ has quit [Client Quit]
<andrewrk>
as tempting as it is, I won't derail the topic with a discussion about build systems, although you are welcome to start such a conversation in #zig of course
___nick___ has joined #ffmpeg-devel
<Sean_McG>
was not my intention, I was just blowing off steam :)
<andrewrk>
kierank: I'm afraid I have never been as valuable a contributor as that to ffmpeg
<cone-795>
ffmpeg Marton Balint release/6.1:88127b743f33: avfilter/af_channelmap: fix error message if FL source channel was missing
<cone-795>
ffmpeg Marton Balint release/6.1:0deb010ae7ee: avfilter/af_channelmap: fix mapping if in_channel was a string but out_channel was not specified
<cone-795>
ffmpeg Marton Balint release/6.1:8d1e092b245f: avfilter/af_channelmap: disallow channel index 64
<cone-795>
ffmpeg Marton Balint release/6.1:aa5e6017a58e: avfilter/buffersrc: fix overriding unknown channel layouts with negotiated one
<compn>
no one singing the herald of pkg-config i see
<Sean_McG>
thank $DEITY
j45_ has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
j45 has quit [Ping timeout: 260 seconds]
j45_ is now known as j45
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
___nick___ has quit [Remote host closed the connection]
<cone-795>
ffmpeg Marton Balint release/6.0:ea2dcb9b3918: avfilter/af_channelmap: fix error message if FL source channel was missing
<cone-795>
ffmpeg Marton Balint release/6.0:1ad767bf2749: avfilter/af_channelmap: fix mapping if in_channel was a string but out_channel was not specified
<cone-795>
ffmpeg Marton Balint release/6.0:fc8775ae9661: avfilter/af_channelmap: disallow channel index 64
<cone-795>
ffmpeg Marton Balint release/6.0:aae59be4001a: avfilter/buffersrc: fix overriding unknown channel layouts with negotiated one
<cone-795>
ffmpeg Marton Balint release/5.1:66430bf110b2: avfilter/af_channelmap: fix error message if FL source channel was missing
<cone-795>
ffmpeg Marton Balint release/5.1:59dccea6ef8f: avfilter/af_channelmap: fix mapping if in_channel was a string but out_channel was not specified
<cone-795>
ffmpeg Marton Balint release/5.1:19c3f055b2e2: avfilter/af_channelmap: disallow channel index 64
<cone-795>
ffmpeg Marton Balint release/5.1:defa085fc88d: avfilter/buffersrc: fix overriding unknown channel layouts with negotiated one
darkapex has quit [Remote host closed the connection]