LarstiQ has quit [Quit: ZNC - http://znc.in]
LarstiQ has joined #hpy
dalley has joined #hpy
<antocuni> hi! I finally have some HPy time this week
<antocuni> any PR which you would like me to review? (I don't want to spend all my time on reviewing though :))
<mattip> antocuni: there is one comment unresolved in https://github.com/hpyproject/hpy/pull/371
<mattip> and then there is the milestone for the next release
<mattip> fangerer: could you rebase the graal-team/numpy-hpy-port off master so we can see what else is left to merge back?
<fangerer> antocuni: yes, absolutely; please re-review my trace mode PR https://github.com/hpyproject/hpy/pull/339
<fangerer> it would also be great if you would finish your hybrid ABI PR (https://github.com/hpyproject/hpy/pull/371)
<fangerer> (as Matti already commented)
<fangerer> mattip: okay, I'll rebase but will push it as a new branch because otherweise the related branches in `numpy-hpy` and `hpy` would be out of sync
<mattip> cool. Although I think that if rebasing it breaks the numpy-hpy build then something is wrong
<mattip> Maybe it is sufficient to look at the commit history
<mattip> to see the big pieces that are still not merged back to master:
<mattip> like structseq, HPy_Type, and more
<antocuni> cool, I'll do both then :)
<fangerer> mattip: TBH, I expect breaks because we often just quickly added APIs/features to HPy as we needed them. Those need to be designed and implemented properly if moved to master.
<fangerer> > like structseq, HPy_Type, and more
<fangerer> As I announced, I'm working on exactly that.
<fangerer> Currently, I'm working on https://github.com/hpyproject/hpy/issues/310 (will open a PR soon). That's not directly required for numpy-hpy but dramatically improves test time which we probably need for Antonio's hybrid ABI PR (since that's adding another dimension of test modes)
<fangerer> I will then work on vectorcall support (i.e. being able to specify `HPy_tp_vectorcall(_offset)` in type specs). That's required for numpy-hpy.
<mattip> thanks