Since the top-level table is getting undefined, all the things in it will
be unreachable and eligible for garbage collection without explicitly
nulling them out.
(cherry picked from commit 5191b20541)
Now, if the AudioContext starts in a "suspended" state, because the browser
blocked it from playing by default, we we run the audio "thread" in a timer
and throw away the generated audio. Once the AudioContext is allowed to
resume, we clear this timer.
The end result is that the app will continue to drain its audio queue
instead of consuming more memory over time (and, if it relies on an audio
callback to make progress, continue to run!), with the effect that the
page is merely silent but otherwise functioning as intended.
Once the user interacts with the page and the browser permits the the
AudioContext to run for real, audio should still be in sync, instead of
just starting to play audio that might now be at least several seconds behind.
(cherry picked from commit fd75a4ca05)
- Make sure the hotplug thread has hit its main loop before letting
DetectDevices continue.
- Don't unref the context subscription operation until it completes
(or we are shutting down).
I'm not sure which change fixed the problem, but at least one of them
appears to have done so.
Reference Issue #7971.
This risks blocking the thread if disaster ensues, and we can wait in the
thread's main loop for subscription as well anywhere else.
Reference Issue #7971.
(cherry picked from commit 956b18f50c)
This used to create a context and mainloop for each device and the hotplug
thread, but this isn't correct use of PulseAudio's API. Now we have a
single context and a pa_threaded_mainloop, and all threads cooperate around
it.
This was originally from SDL3, in 35292d7dba.
Reference Issue #7883.
Reference Issue #7427.
This prevents the case where the mouse might be at the edge of the
window when enabling relative mode, which confuses macOS, at it
might believe the user is attempting to resize the window.
Fixes#6994.
They can vanish for UP TO EIGHT SECONDS...!
This is for devices that connect to HDMI/DisplayPort/etc, where it
presumably has to wait for a display to get up and running before it
can play audio through it, so one can see the audio device fail when
changing display modes, or the system returning from sleep. Since this
can be triggered by a game changing video resolutions at startup (either
before or after opening the audio device!), it's important to deal with.
In normal conditions, it shouldn't take this long to open or recover an
audio device, but this is better than unexpectedly losing the device
in this situation.
Fixes#7044.
Fixes#5571.
In theory this is illegal, but legit wavefiles in the field do it, and
it's easy to bump it to 1 for general purposes.
Formats with more specific alignment requirements already check for them
separately.
Fixes#7714.
(cherry picked from commit 2e646c7141)
sdl12-compat can get into a state where a color-keyed surface is
marked for blending, but wants to blend with full alpha (which
is the same as _not_ blending), so rather than fail to find a
blitter in that case, it just selects the colorkey blitter.
Reference https://github.com/libsdl-org/sdl12-compat/issues/233
This adds a means to mass-convert the whole wiki to Markdown as a one-time
operation, and then some fixes to make --copy-to-headers correctly deal with
Markdown-formatted wiki pages.
(cherry picked from commit 936a51d5cc)
We get audio artifacts if we don't work at the higher precision, but
this is painful on CPUs that have to use a software fallback for this,
so for now (that is, until we have a better solution), get better output
on amd64 chips, where the cost is less painful.
(cherry picked from commit 1e5e8e2fda)
This needs to check what our deployment target is, not what SDK
is available, since this is a linker symbol and not an enum
value or whatever.
Also removed a copy/paste error that mentioned CoreAudio in
the haptic subsystem.
Fixes#6534.
This reverts commit 04b50f6c6b.
It turns out OpenGL vsync has broken again in macOS 12, so
we're reintroducing our CVDisplayLink code to deal with it,
again.
Reference Issue #4918.
This uses a newer browser API to get physical scancodes, but still
uses the (deprecated) event field that we were already using for
scancodes, but for keycodes instead now, which appears to be more
accurate.
Since keyboard layout isn't (generally) available to web apps, this
adds an internal interface to send key events with both scancode
and keycode to SDL's internals, instead of sending just scancodes and
expecting SDL to use its own keymap to generate keycodes.
Future work in this area would be to use the keyboard layout APIs
on browsers that support them, which would allow us to use SDL's
usual keymap code and not rely on a deprecated browser API, but
until we get there, this patch gives significantly more correct
results than we would have before.
Fixes#2098.
This ignores 2.x.1 (etc) releases, which prevents it from thinking
the next official non-point-release version is 2.26.1, when it
should be 2.26.0, because it saw the "latest" release is 2.24.1.
This fixes the wiki ending up with imaginary version numbers for
the "this function is available since SDL 2.x.y" sections.
Fixes#6343.
Otherwise the thread might block for a long time (more than 10 seconds!).
It's not clear to me why this happens, or why its safe to do this with a
resource that's still in use, but we have, until recently, always
disposed of the AudioQueue first, so changing back is probably okay.
Also changed the disposal to allow in-flight buffers to reach hardware;
otherwise you lose the last little bit of audio that's already been queued
but not played, which you can hear clearly in the loopwave test program.
Fixes#6377.
This will only log things going through dynapi, which means it won't
do anything if dynapi is disabled for a given build, but also things
that call the `*_REAL` version of an API won't log either (which is
to say, if an internal piece of SDL calls a public API, it won't log
it, but if an application calls that same entry point, it will).
Since this just inserts a different function pointer, unless you
explicitly request this at runtime, it won't add any overhead, and,
of course, the entire thing can be turned off with a single #define
so it doesn't even add extra unused code to the shared library if
the kill switch is flipped.
Ryan C. Gordon