<mattip> there are no expectations from the HPyThreadState? No c-extensions expect certain fields in it?
<antocuni> mattip: exactly
<antocuni> this is the definition of the HPyThreadState struct from the point of view of extensions:
<antocuni> as you can see, it's completely opaque
<mattip> it is hard to see how that will be useful
<mattip> the functions could have only acquired/released the GIL , with no thread state needed
<mattip> in the graal-team/hpy branch of numpy
<mattip> they have "HPyThreadState _hpy_token" but do nothing with it
<mattip> so I guess it is up to the python implementation to add meaning to that token?
<mattip> The documentation does not shed light on the use of HPyThreadState
<antocuni> re "I guess it is up to the python implementation to add meaning to that token?"
<antocuni> yes
<antocuni> I think that the main reason why it's there is that it's needed by CPython
<antocuni> it's perfectly fine to ignore it in PyPy if you don't need
<antocuni> mattip: I suggest to keep all the conversation about hpy in this channel, else it becomes hard to follow in two channels :)
<mattip> ok, sorry.
<antocuni> no problem :)
<cfbolz> ok, maybe the graalpy people can complain if things become too pypy specific
<antocuni> that's true, but on the other hand I thought it might be interesting to hear their opinion/advice on some aspects
<antocuni> e.g. how the implemented HPyGlobal
<antocuni> *they
<antocuni> but of course, if they prefer us to talk in #pypy, just let us know