PEP 788 -- Protecting the C API from Interpreter Finalization

[ZeroIntensity]

Please consider PEP 788 -- Protecting the C API from Interpreter Finalization
https://peps.python.org/pep-0788/

This PEP was approved by the C API Working Group: capi-workgroup/decisions#70 (comment)

• The PEP has been discussed in threads listed in its Post-History header
• The PEP was announced on Discuss (link in Post-History)
• The PEP includes all relevant Suggested Sections
• The PEP includes endorsements from the projects/groups/people it helps
• The PEP has a CODEOWNERS entry

SIG-specific:

• Typing PEPs: link to the Typing Council's recommendation for the PEP:
• Packaging PEPs: don't file the issue here, ask the delegate (Paul Moore) on Packaging Discourse

[NicolasT]

Nit: I think there's a minor bug in the example code in the PEP, passing a const char * returned by PyUnicode_AsUTF8 to free.

[ZeroIntensity]

Yeah, initially that wasn't a call to PyUnicode_AsUTF8 and I just forgot to remove the free call when I changed it. Good catch!

[corona10]

Hi @ZeroIntensity and @vstinner

SC replied comment on capi-workgroup/decisions#70 (comment) for this PEP.
Please take a look :)

Warm regards,
Donghee

[corona10]

Hi @ZeroIntensity,

Last week we discussed PEP 788, and we realized that it would be helpful to have an Office Hour with you to better understand the details.
If you don’t mind, would you be willing to request an OH?

Warm regards,
Donghee

[ZeroIntensity]

Hi Donghee,

Thanks for the update. I've scheduled office hours for March 19th.

[ZeroIntensity]

Hi SC,

I've updated PEP 788 based on our discussion from a few weeks ago.

Notably:

1. PyGILState is now only soft deprecated by the proposal (i.e., we continue to ensure it works and is supported, but we won't add new APIs for it).
2. PyUnstable_InterpreterView_FromDefault is now PyInterpreterView_FromMain and is added to the limited API.
3. No more PyThreadView or opaque integer types (PyInterpreterGuard is now PyInterpreterGuard *).
4. Added PyThreadState_EnsureFromView.
5. Removed a number of APIs that were there for completeness (such as the _Copy functions).

I also refined the "Motivation" and "Rationale" sections to make the PEP's goals a little clearer.

Combined, these changes should hopefully address the concerns regarding the PEP's scope and size. Subinterpreter support is now very easy to opt out of (via PyInterpreterView_FromMain), and existing PyGILState code will continue to work.