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
Gramner has quit [Remote host closed the connection]
<nevcairiel>
if you have tested cl.exe and clang-cl with the change, then its just fine
<mkver>
wbs: Does MSVC actually support _Alignas? If so, we could remove the MSVC specific case.
<nevcairiel>
in theory it supports it, but there was some breakge when the original patch tried to use it
<nevcairiel>
hence why the MSVC case was kept
<wbs>
mkver: it was attempted to use the C11 alignas() in 57861911a34e1c33796be97f2b2f44e05fffd647, but it showed some issues in MSVC, so an MSVC specific case was readded in 6e49b8699657b808b7dc80033f2c3f2d0e029fa3
<mkver>
I thought the breakage was that MSVC (at least some versions of it) does not provide stdalign.h, so if we used _Alignas instead of alignas, it might work.
<wbs>
not sure exactly what the breakage was, 6e49b8699657b808b7dc80033f2c3f2d0e029fa3 is light on details
<nevcairiel>
it wasnt a compile error, but some fate tests broke
<nevcairiel>
it seems to miscompile the alignment in some specific cases
<wbs>
anyway, that's orthogonal, I'll go ahead and ack this on the ML and push the fix, so I can set up fate instances for clang/msvc on x86 (I already have that on arm, but they missed this issue)
cone-040 has joined #ffmpeg-devel
<cone-040>
ffmpeg Martin Storsjö master:fb65ecbc9b80: avutil: Fix linking x86 asm constants with Clang in MSVC mode
mkver has quit [Ping timeout: 252 seconds]
Everything has joined #ffmpeg-devel
meego has quit [Remote host closed the connection]
IndecisiveTurtle has quit [Ping timeout: 265 seconds]
rvalue has quit [Ping timeout: 272 seconds]
microlappy has joined #ffmpeg-devel
microlappy has quit [Client Quit]
Lynne has quit [Read error: Connection reset by peer]
meego has quit [Remote host closed the connection]
rvalue has joined #ffmpeg-devel
cone-040 has quit [Quit: transmission timeout]
<kasper93>
wbs: the patch looks ok and it;s unfortunate that we need specific _MSC_VER codepath for cl.exe
<kasper93>
I'm lagging behind with testing master version, because I prefer meson buildsystem for MSVC-like builds. ./configure is to flakely in this configuration, you need to align all the stars exactly to make it work...
<wbs>
ah, that explains why you hadn't run into it
<wbs>
yeah the meson wrapping of ffmpeg is quite neat
<kasper93>
I tried yesterday quickly to build it, but I eventually run into `LINK : fatal error LNK1181: cannot open input file 'm.lib'` and didn't look further
<wbs>
it works quite ok for me (although I mostly cross compile from linux); in a terminal with MSVC environment variables, start msys2 with the -full-path (or what it's called) option, to keep the MSVC path entries in place within msys2; then just "./configure --toolchain=msvc" should be enough
<kasper93>
maybe I should set up GHA jobs for my configuration
<BtbN>
configure on MSVC works fine for me
<BtbN>
running it almost daily for the fate builds
<wbs>
oh right sorry that's for plain msvc builds; building with clang in msvc mode probably requires manually specifying a couple more flas
<wbs>
*flags
<kasper93>
I know it can work, it just is not streamlined
<BtbN>
It definitely has issues though, like partial rebuilds on updates not working
<BtbN>
But it's not meant as a prime development target anyway, just as oneshot and forget builds, and it does that just fine
<kasper93>
If I get some time after $dayjob, I will look at it
jamrial has joined #ffmpeg-devel
Lynne has joined #ffmpeg-devel
<compn>
OctopusET, ask j-b , maybe in #videolan
<OctopusET>
okay thannks
<compn>
OctopusET, you are ok with traveling to europe? its usually near france because a lot of developers are in that country
<OctopusET>
sure
<OctopusET>
Or Oui
System_Error has joined #ffmpeg-devel
minimal has joined #ffmpeg-devel
minimal has quit [Remote host closed the connection]
Flat_ has joined #ffmpeg-devel
minimal has joined #ffmpeg-devel
Flat has quit [Ping timeout: 272 seconds]
lexano has quit [Remote host closed the connection]
lexano has joined #ffmpeg-devel
<wbs>
kasper93: oh, also, if you build with --cc=clang instead of --cc=clang-cl, you may need --extra-ldflags='msvcrt.lib oldnames.lib'; the embedded directives don't always get added when using the "clang" interface in the same way as they get added with "clang-cl"
<kasper93>
yeah, I see that
<kasper93>
tried also --ld=clang --extra-ldflags=-fuse-ld=lld, but this fails in another ways
<kasper93>
btw, sed: can't read /e/dev/ffmpeg/fftoolsresources/graph.css: No such file or directory
<kasper93>
nice path joining
fflogger has quit [Ping timeout: 252 seconds]
fflogger has joined #ffmpeg-devel
fflogger has quit [Ping timeout: 272 seconds]
fflogger has joined #ffmpeg-devel
Kei_N_ has joined #ffmpeg-devel
Kei_N has quit [Ping timeout: 272 seconds]
minimal has quit [Quit: Leaving]
fflogger has quit [Ping timeout: 272 seconds]
fflogger has joined #ffmpeg-devel
<kierank>
thardin: "If ffmpeg could do NLE, that would not be a bad thing"
<kierank>
You have to understand several people in this project want ffmpeg to be the kitchen sink of multimeia
<kierank>
so don't give them ideas
fflogger has quit [Ping timeout: 248 seconds]
<Lynne>
improve the NLE in blender instead pls
<TheVibeCoder>
improve the multimedia in gstreamer instead pls
<thardin>
kierank: yes, I'm aware
<thardin>
we have to deal with elst though. I don't think removing elst handling would fly
meego_ has joined #ffmpeg-devel
<thardin>
exposing it as side data seems reasonable enough, and is useful also for mxf and imf
<thardin>
personally I'm in favor of changing behavior, so that the people who are adamant about ffmpeg rendering things according to elst come forward and implement things properly
<thardin>
I was tempted to suggest that, with a sufficiently generic solution, we could even implement AAF if we really wanted to. but that feels way out of scope. MXF is bad enough
<thardin>
haha time for moose ramen
<averne>
Lynne: I'd like to integrate renderdoc captures in the vulkan code to ease debugging, do you have feelings about this? (basically is this something I should do cleanly and try to upstream or should I just hack it in)
<Lynne>
averne: IndecisiveTurtle already had some patches
<averne>
Oh ok, I'll try to find these and pull them in my local tree
<averne>
Couldn't find them, I'll just do it myself
<averne>
So is this something you'd want in ffmpeg master?
<Lynne>
the patches he had were messy
meego_ has quit [Ping timeout: 260 seconds]
<thardin>
oh yeah before I forget: is enough interested in RE work using IDA?
<thardin>
a friend of mine knows a company that is hiring
<TheVibeCoder>
no Ghidra?
<thardin>
hexrays ida is what my friend said
<thardin>
but maybe they're not dead set on that
<TheVibeCoder>
hope they pay full IDA license
<thardin>
no idea
bsFFFFFF has quit [Ping timeout: 252 seconds]
<kasper93>
trac is down or it's just me
<kasper93>
?
mkver has joined #ffmpeg-devel
Guest5 has joined #ffmpeg-devel
Guest71 has joined #ffmpeg-devel
Guest5 has quit [Client Quit]
Guest71 has quit [Client Quit]
<Traneptora>
kasper93: loading here
ngaullier has quit [Remote host closed the connection]
DauntlessOne49 has quit [Remote host closed the connection]
DauntlessOne49 has joined #ffmpeg-devel
Guest43 has joined #ffmpeg-devel
Guest43 has quit [Client Quit]
IndecisiveTurtle has joined #ffmpeg-devel
minimal has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 244 seconds]
Everything has quit [Quit: leaving]
Kei_N has joined #ffmpeg-devel
Guest48 has joined #ffmpeg-devel
Kei_N_ has quit [Read error: Connection reset by peer]