Closed KKQ-KKQ closed 8 months ago
2 questions:
- If we move to having a single background thread for loading, this logic and the atomics would get much easier, wouldn't it?
If we moved to having a single background thread for loading and having shared file pool, an instance have to wait until another instance finishes loading.
- It seems we're so close to not needing the
loadedFiles
. The only issue I see is if someone callssetRamLoading(true)
, and thensetRamLoading(false)
, which would erase theFullyLoaded
states from before.
loadedFiles
is for smaller files and for wave-oscillators and the samples in it should not be released even when preloadSize
is changed.
I think we needs the loadedFiles
.
2 questions:
- If we move to having a single background thread for loading, this logic and the atomics would get much easier, wouldn't it?
Oh do you mean finalizeSfzLoad()
just queue loading? I think it might be more complex, but worth consider it.
I think the processes become more safe.
Looks good, I'll try to stress test it a bit today or tomorrow thanks!
@KKQ-KKQ I'm sorry I've been way too swamped at work. I'm merging this without extensive testing so you can proceed. I'll test out things as they come when I can...
すみません ! 🙇
@KKQ-KKQ I'm sorry I've been way too swamped at work. I'm merging this without extensive testing so you can proceed. I'll test out things as they come when I can...
すみません ! 🙇
No problem! Thank you!
PR 1/4 of #1236 Shared File Pool
Added FileData::Status::FullLoaded.