To GIL or not to GIL

[McSinyx]

As described in Cython's documentation, the Python GIL may need some special care. Currently all C++ classes are declared with nogil C++ routines are randomly marked with nogil and with gil, and I am unsure whether it is not a good practice.

Edit: the acquire and release of GIL are definitely problematic, and so far all discovered deadlocks are with resource loading:

1. Access to a buffer being cached, either via Buffer or free (Deadlock if Buffer called on the resource being cached async #73).
2. Streaming a decoder defined in Python (deriving from BaseDecoder). Note that if it's loaded into a Buffer, the caching process is done on the main thread and thus no deadlock occurs.
3. Loading resources using custom FileIO, either as a stream (decode) or as a buffer. This can either be a deadlock or a segfault.

[McSinyx]

A minimal example to reproduce would be

from palace import *

use_fileio(lambda name: open(name, 'rb'))
with Device() as dev, Context(dev) as ctx, decode('audio file').play(12345, 6) as src:
    while src.playing: ctx.update()

[McSinyx]

I'd be grateful if you two @Huy-Ngo and @9a24f0 can do a little bit of research on this and try to solve it. If none of us figure out a fix by the end of avril, we can declare this as a technical debt and precede to release v0.2 anyway.

Edit: it's the end of the month, this is dropped from the list of 0.2-blocking issues.

[McSinyx]

Since 1.0 is supposed to be the stable version, this should be fixed then.

[Huy-Ngo]

C++ routines are randomly marked with nogil and with gil

I'm a bit confused here. Which routines are marked with with gil? As I can see, none in alure.pxd is marked with with gil. Do you mean the ones in Cython's implementation?

[McSinyx]

There are C++ functions defined in palace.pyx too (mostly for interfacing, e.g. Decoder, MessageHandler, FileIO).

[McSinyx]

Migrated to SourceHut TODO ~cnx/palace#1.