JuliaPy / JuliaPy/PythonCall.jl

JIT Python option and other options to take advantage of

Aperta
#496 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub
enhancement
Lingua principale
Julia
Stelle
1.1k
Fork
86
Merge medio
1g 22h
PR unite (30g)
3

Descrizione

I see you upgraded minimum Python to 3.8, so it got me curious what does it and later versions enable that you could take advantage of:

https://docs.python.org/3.13/whatsnew/3.13.html#experimental-jit-compiler

You download Python automatically, and while the JIT is in the current alpha, it's experimental and not on it seems by default.

I.e. I think it requires a specialized build, and you could support such, optionally. Or by default, at least it will be at some point. Also they are dropping 19 outdated batteries, so it's a smaller download...

In 3.8:
https://docs.python.org/3.8/whatsnew/3.8.html#parallel-filesystem-cache-for-compiled-bytecode-files

https://bugs.python.org/issue33499

Yes, it's an ENV var and your users could just set it, or be ignorant of it. I was thinking should you set it for the users? That might be bad in case the user already set it... Anyway, it could be documented here and/or set for the user, unless you invoke PythonCall in a special way to not disturb it.

I didn't read carefully, but it seems can be much faster (startup) if you opt into it, I can understand if you do not want to change defaults for users, or document to much. Maybe just document 3.13 is preferred, but older down to 3.8 is supported? With some easy way to limit to 3.13 and newer.

Guida per i contributori

Nessuna guida per i contributori indicizzata per questo repository

Valutazione

Questa issue non è ancora stata valutata.

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.