03:56 <mattip> antocuni: about HPy_GLobal_Store/Load: if I understand correctly we need to keep alive the global index of the handle and any app-level objects it might wrap

03:56 <mattip> do we need to set aside a range of indices that will be used specifically for global handles in the handle manager?

03:57 <mattip> is it correct that the logic must manage the lifetime of the wrapped objects separately from the logic in handlemanager?

08:14 <antocuni> mattip: yes, that's correct. Also, I think that the idea is that they are automatically per-thread, but better to ask fangerer and steve_s to double check

08:15 <antocuni> but I think that for now it's perfectly fine to store them as a field in the State (and think about threading later) is we are not sure

08:16 <fangerer> We can leave it at that for now. We were just discussing that the design of HPyGlobal also allows to be used with subinterpreters.

08:17 <fangerer> antocuni: since you are obviously online 🙂 : if you find the time, please do a first review round for https://github.com/hpyproject/hpy/pull/364

08:18 <antocuni> ahah you caught me 😂

08:18 <antocuni> yes, it was on my TODO list for today

08:18 <antocuni> in Germany it's a holiday but I'm happy to dedicate some time to HPy today

08:20 <fangerer> 🤣

08:20 <fangerer> sorry; I didn't know that there is a public holiday in Germany today

08:23 <antocuni> no problem, really

10:16 <fangerer> antocuni: thanks for reviewing; will address your comments ASAP.

10:16 <fangerer> Since you are already in HPy: here is another tiny but I think useful PR. Should be very quick to review: https://github.com/hpyproject/hpy/pull/365

11:02 <antocuni> done :)

12:19 <fangerer> thank you

16:43 <steve_s> > I think that the idea is that they are automatically per-thread

16:43 <steve_s> If that is meant per Python thread, then no that wasn't my intention. My intention was per (sub)interpreter.

18:36 <mattip> ahh, there is quite a bit of design philosophy in the documentation of the global store, I should have read that before :)

18:36 <mattip> https://docs.hpyproject.org/en/latest/api-reference/hpy-global.html?highlight=x1#c.HPyGlobal_Store

20:39 <mattip> I added a global store to the thread-local ExecutionContext in PyPy in 73d93f41263b, it seems to work as advertised

20:40 <mattip> is the fact that it is thread-local going to cause incompatibilities?

20:40 <mattip> I also added some GIL handleing in later commits

20:41 <mattip> The tests pass, but something is off after translation when the JIT kicks in

20:42 <mattip> and the tests hit HPy_LeavePythonExecution

20:42 <mattip> s/The tests/Untranslated tests/