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 7.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
<BBB>
michaelni: re: "vf_colorspace: Add an option to clamp trc LUT output", I reviewed that in another thread also
<BBB>
michaelni: it's a bit annoying but there's 3 versions of that patch on the ML and it's impossible to link them in any meaningful way in mailboxes :(
realies9346 has quit [Quit: ~]
minimal has joined #ffmpeg-devel
linkmauve has left #ffmpeg-devel [not in this room]
<fflogger>
[editedticket] mjuhasz: Ticket #7756 ([avcodec] add mastering display color volume and content light level information to hevc_metadata bitstream filter) updated https://trac.ffmpeg.org/ticket/7756#comment:2
j45 has quit [Ping timeout: 276 seconds]
j45_ is now known as j45
j45 has joined #ffmpeg-devel
j45 has quit [Changing host]
<JEEB>
Lynne: thanks for the review on that dull patch set :) btw do you recall if compilers actually now error out if you try to utilize a variable that is not defined yet? since I recall a few years back in mpv we had a case where we had a variable definition in the middle of a function and when someone added a usage of that variable before that definition it happily compiled even though the variable was
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 8.0 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
<JEEB>
Traneptora: good. I wish I recalled more details regarding what the case was where the variable was happily accessed or set before being defined
<Traneptora>
JEEB: iirc the issue was that it was defined with an initialier, like AVFoo *bar = NULL; and then we called free on it after an end: label
<ePirat>
JEEB, Traneptora, dont compilers warn if goto skips over variable declarations/init?
<JEEB>
at least then it did not
<JEEB>
you'd hope/wish they did
mkver has joined #ffmpeg-devel
<Lynne>
michaelni: ...
<Lynne>
I did ask for a couple of hours of warning
<Lynne>
though I'd been fine if it was 30 minutes or even 3 minutes
<BtbN>
Does it matter that much that the website post is there the instant of the tag?
<Lynne>
not really
<Lynne>
its all pointless
<BtbN>
As long as the post isn't there before the tag, it doesn't matter too much imo
<BtbN>
plenty of projects even have days between technical release and announcement, to have archives propagate to mirrors
<Lynne>
no, I mean it
<Lynne>
there's no point in posting anything, no one reads it
<BtbN>
Plenty of people read the announcements, what
<Lynne>
everyone copies phoronix, who know next to nothing
<JEEB>
hah, it seems like clang's scan-build actually found a valid issue
<BtbN>
32bit fate CI is broken btw.
<BtbN>
it found an issue near instantly it seems
<JEEB>
not a major one (value is stored into function scope variable, but then never read as the value is only overwritten and not read from until the end of function
<JEEB>
BtbN: yea, aacenc checkasm apparently
<wbs>
if the fate log doesn't print the actual .err file, we don't know the seed used for the test, so it's hard to reproduce...
<JEEB>
yea
<JEEB>
I was going to say that could be a good improvement for the script, to grab the message and cat that mentioned file
<wbs>
then again, it took me all of 1 second to hit a failing seed for that one
<wbs>
i=0; while true; do ./tests/checkasm/checkasm --test=aacencdsp $i || break; i=$(($i+1)); done
<JEEB>
:D
<wbs>
fails for seed 41
<JEEB>
almost the answer
Anthony_ZO has quit [Remote host closed the connection]
marcj has quit [Ping timeout: 248 seconds]
marcj has joined #ffmpeg-devel
<kasper93>
bruh, it's already on phoronix
<kasper93>
there was not even official news page update
<kasper93>
> with significant performance benefits for AVX-512 CPUs
<kasper93>
in filter that you will never use
<kasper93>
ever...
<galad>
well, it's phoronix, if you sum all the performance improvements in their article titles linux should fly on a 10 mhz cpu
<michaelni>
Lynne, i did say it before going to bed yesterday "<michaelni> Will do the 8.0 release tomorrow, dont want to rush it now before i go to bed "
LuisF has quit [Quit: Client closed]
<Lynne>
haasn: honestly, I dumbed down the text as much as possible already, what excessive technical details do you see?
<Lynne>
its already the bare minimum to avoid misunderstandings
<Lynne>
michaelni: I've pushed the announcement, after making sure there are no typos, checking allcodecs/allformats for a complete list, and addressing all issues from the ML
<Lynne>
kierank: could you announce it on twitter
mkver has quit [Ping timeout: 248 seconds]
mkver has joined #ffmpeg-devel
<BtbN>
The migration for ffmpeg-user seems to have worked fine enough
<BtbN>
I've tried to google for "MIPS Servers"
<BtbN>
and my god, it's impossible to find anything sensible
<BtbN>
apparently there's a German Hosting-Company called mips
<sfan5>
why would you buy servers for a dead cpu architecture
<haasn>
Lynne: I don't like that you spent about 70% of the words in the release notes on a feature that is less than 10% of the diffstat between 7.1 and 8.0
<haasn>
who cares about the technical details of what codecs are candidates for some obscure vulkan compute rewrite nobody will care about
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<Lynne>
even if they don't use it, its something interesting
<Lynne>
users request GPU accelerated codecs all the time, here's a couple
<Lynne>
if you asked me to, I could've written 10x the amount on just the prepwork you did for the swscale rewrite
<GewoonLeon>
is there like some rule against using bools or why are ints so often used for booleans? is it just to cut down on #includes?
<JEEB>
stdbool was not available in earlier versions of C
<JEEB>
now it can be utilized internally
<GewoonLeon>
aight, i see
<BtbN>
a bool is also just an int iirc?
<BtbN>
so not really any advantage
<frankplow>
I think a bool is more typically a char than an int
<frankplow>
But the value of true is defined in the spec to be equal to 1 and false to 0, so you don't get any additional type safety or anything just expressiveness