For VP9, libsvt_vp9 is preferrd (if built-in, which is in Linux builds),
but it compiles with AVX2 by default, so prefer libvpx if CPU doesn't
support that.
PortAudio/ASIO drivers seem refuse initialization if
CoInitilizeEx initialized as COINIT_MULTITHREADED but they do for
COINIT_APARTMENTTHREADED.
Windows devices that could be affected by this change tested and seem
to work, namely:
- DeckLink
- DirectShow
- WASAPI
refer to GH-430
The commit b785c3d8 from 2024-08-05 changed the default unconditionally
to PCM, but this has been true just for UltraGrid RTP, not RTSP or SDP
that used MP3, so this change reverts it back.
If the input is already compressed, try to pass it further. Initialize
the compression just if receiving uncompressed data.
This is implemented in the same way as it already is in vrxtx/sdp.
refers to GH-433
video_display putf implementation may free the frame so that the pointer
is no longer valid upon the exit from the callback (or anytime later).
This is the case for eg. vdisp/decklink.
Fixes `uv -t testcard -c lavc -d dummy:discard` (wrong color in output
line):
```
[dummy] 126 frames in 5.03501 seconds = 25.0248 FPS
```
even though FPS is +/- the nominal 25. Depending on circumstances it
could have also easily crashed (access after free).
The mutibyte to wide character conversion was (perhaps since the
beginning) wrong - mbtowc converts just one character - it should have
beem mbstowcs but using rather mbsrttowcs (thread-safe).
There are some calls using just the codec_ctx
parameter, which fail (with the compat with
legacy FFmpeg).
eg. `uv -A Opus -s testcard` (using the old FFmpeg)
1. static_assert could not have been strictly speaking used (splash_width
is not constexpr), although it worked most of the time (except macports
clang-mp-14)
2. the assertion was actually inverted (but the equality is included
and is the case for both cases so both work)
select the color automatically (lime for +- 2-5%, brown for >5% diff)
This simplifies the code by sharing the logic in common; also exceeding
FPS is now colored (until this just lower than nominal was colored).
If the actual number of decoded samples doesn't match the expected number
(from nominal sampling rate), use colors to catch user attention. This is
similar how it is already done for video capture/display.
In the display generic FPS indicator, use colors (arctic lime/saddle
brown) if the actual FPS is lagging behind the nominal FPS. Similarly
(and copied from) how it is already done for video capture.
The pitch for planar pixel formats is the pitch of the first plane,
most likely =width. This yield wrong data_len for I420 if just multiplied
by the height.
this fixes the change made by 107e3e30 (2024-09-02)
include inttypes.h
SDL_CHECK can now have an optional action (using ellipsis) + handle
return value of SDL_LockTesxture using that (currently fails on Arch
Linux with sdl2-compat 2.30.50-1 and sdl3 3.2.0-1 and YCbCr textures).
v0.26 is slightly more chatty even in GPUJPEG_LL_STATUS mode (new between
INFO and VERBOSE) while the status/verbose mode maps more to debug than
UG verbose so adjust the modes as such.
allows build outside of UG/when config.h not generated
Also helps when building tools/convert even when config.h is generated,
VDPAU enabled but FFmpeg headers in non-standard path (that is obtained
by pkgconf for UG but not for the tool).
No longer enforce v4 socket for dot-decimal IPv4 mcast addresses unless
macOS used with mcast4 addr and interface set.
The default behavior now is to use dual-home v6 socket (with v4-mapped
addresses). macOS treates the v4 mcast addresses transparently as
v6 addresses (using v6 sockopts). But interface specification doesn't
currently seem to work in macOS, neither for native v6 mcast addresses
so enforce v4 sockets here.
- support for v4-mapped IPv6 sockets (handled with v4 sockopts in
Linux+Win and v6 sockopts in macOS)
- for mcast4, except of Windows, the interface must be identified by
local address, not device number
- support also setting the bind address directly for v4
- leave mcast4 group with correct ID (not INADDR_ANY)
The local address interface specification was actually the original one
but was removed by the commit 92e24dd02 (in 2012).
macOS fails on it otherwise (at least 15.0), although it shouldn't
according to man ip6(4) (that is, however, presumably inherited from BSD,
because FreeBSD has the same).
This fixes the commit 43d15680 (2021-03-29) that added this for both
IPv4 and IPv6 but for IPv6 the condition was incorrectly aligned to
setsockopt fail, not ttl==-1.
Dropped IPv4/IPv6 specification - since we may/will use v4-mapped IPv6
sockets, the word "IPv6" might be confusing if use is using IPv4 mcast
addr.
Since the following error, the hard fail is again desirable. It, however
doesn't fail if the driver is usable (not so old, using same ABI).
This reverts commit cac114d6e5.
