<antocuni>
> is the fact that it is thread-local going to cause incompatibilities?
<antocuni>
I think so. According to what steve_s says they are not thread-locals on CPython, which means that as soon as you have two threads, your HPyGlobal_Load would return two different things in CPython and PyPy
<antocuni>
sorry for the having caused the confusion. I vaguely remembered something like "HPyGlobal is not truly like a C global" and I thought it was per-thread, but it's actually per-interpreter
<antocuni>
(which is not a "problem" for pypy since we don't support sub interpreters at all)
AntonioCuni[m] has joined #hpy
AntonioCuni[m] is now known as antocuni[m]
steve_s has quit [Quit: Client closed]
<mattip>
OK, I will find a place to "hang" the globals on the interpreter then, thanks
<mattip>
other than that, I think the hpy-0.0.4 branch is ready
<cfbolz>
mattip: you should be able to store it in the State instance
<cfbolz>
that's per interpreter
<mattip>
+1
<antocuni>
yes, side by side with the normal handles