My understanding is that "pause" was introduced with the Pentium 4 and
assembles to opcode F390 which is the same as "rep; nop". I believe this
should be backwards compatible.
Using "pause" here introduces an compilation error when targeting
architectures older than Pentium 4.
These were undefined when linking with the -static flag on:
1. Alpine Linux v3.21 x86_64
2. Debian GNU/Linux 12 (bookworm) aarch64
In both cases I had to manually compile alsa-lib as a static library
because the systems have only .so versions. I did it with:
cd path/to/alsa-libs
./gitcompile static
<this will create alsa-libs/src/.libs/libasound.a>
Tested with alsa-libs 1.12.2 and 1.12.3.
It's possible to reproduce the errors like this:
clang -static -DMA_NO_RUNTIME_LINKING -DMA_ENABLE_ONLY_SPECIFIC_BACKENDS -DMA_ENABLE_ALSA -DMA_DEBUG_OUTPUT -Wall -Wextra -g3 simple_playback_sine.c -o build/miniwave -fcolor-diagnostics -ferror-limit=2 -fuse-ld=lld -I../
-L/home/antharas/code/alsa-lib/src/.libs/ -lasound
However, even though it compiles I can't verify that everything works.
1. Alpine has musl which doesn't support runtime linking when
compiling with -static. So even though I disabled it for
miniaudio.h, libasound.a itself uses dlopen on
libasound_module_pcm_pipewire.so.
2. Debian uses glibc so it supports it but fails because my machine is
a VPS that doesn't have a soundcard.
No big deal, I don't really need to statically link, it was just
something I stumbled upon.
Also these are currently unused:
_snd_pcm_hw_params_set_rate_minmax
ma_timer_init
ma_timer_get_time_in_seconds
ma_device__on_notification_rerouted
This checks the result of reading from the decoder when asynchronously
loading a file. In practice the existing check already covers this, but
it's technically possible for an error result the be skipped over.
David Reid
Pavel Galkin
Your Name
caturria
znakeeye