10:13 <mattip> dylanyoung: (for the logs) cool!

15:11 <Alex_Gaynor> cfbolz: I just read a blog post on the typed C extensions work, very cool. it should be possible to add support to pyo3 (for rust extensions) as well at some point

15:12 <cfbolz> Alex_Gaynor: oh, that would be fantastic

15:13 <cfbolz> Alex_Gaynor: basically I don't think anyone should do this by hand, but if we could get the right thing to happen for binding generators, that would be amazing

15:13 <Alex_Gaynor> pybind11, which uses C++ templates IIRC, also seems like a good target

15:15 <cfbolz> Alex_Gaynor: right, cool

15:19 <cfbolz> Alex_Gaynor: do you think it's likely that pyo3 would accept a patch like that?

15:21 <Alex_Gaynor> I guess it depends on if it's standardized + how big the patch is. I think if it's non-standard + a huge patch, maybe not. But if there's a PEP and the patch is not too bad (I don't think it'd be big), I think probably yes.

15:22 <cfbolz> Right

15:23 <cfbolz> Alex_Gaynor: it's not clear to me whether there would be buy-in from cpython, given that it would not help them (at least at this stage) at all

15:23 <Alex_Gaynor> CPython is working on a JIT now, maybe they'll care :-)

15:25 <cfbolz> Alex_Gaynor: yep, but unboxing is still quite a while off for them I think

15:25 <Alex_Gaynor> Probably, yes

15:25 <Alex_Gaynor> I suppose for full JIT unboxing you also need a way to declare return types

15:25 <cfbolz> Alex_Gaynor: yep, our prototype does that

15:25 <Alex_Gaynor> ah cool

15:26 <cfbolz> But tons of details to clarify, what about exceptions for example

15:28 <cfbolz> Alex_Gaynor: anyway, in some sense we don't really need cpy buy in if we can agree between, say: pypy, GraalPy, some popular extensions or binding generators. The proposal is binary compatible with cpython after all

16:48 <fijal> "handle_meth" is a fantastic function name