#pypy on 2021-09-01 — irc logs at libera.irclog.whitequark.org

11:17 <mattip> actually, it is now Sept1. We have <200 failures with lib-python, I have done a first draft of a release note (still needs some more things)

11:17 <mattip> https://doc.pypy.org/en/latest/release-v7.3.6.html

11:17 <mattip> so maybe soon should be "very soon"

11:46 <LarstiQ> and then we move on to "Real Soon"?

12:12 <cfbolz> :-)

12:19 gef has joined #pypy

12:22 gef_ has joined #pypy

12:27 <cfbolz> mattip: I wrote a paragraph about the new JIT behaviour in 3.8. It's a bit long, maybe it needs moving somewhere

12:52 <mattip> thanks. Can you suggest an alternative to "Previously, this would prevent compilation of the function and lead to very bad compilation times regardless"?

12:52 <mattip> Perhaps

12:53 <mattip> Previously, this would take a long time to compile and would sometimes even fail to compile. The next time the JIT saw the same code it would try again ...?

12:54 gef__ has joined #pypy

12:54 <mattip> Maybe turn it into a short blog post that then we could link to

12:55 gef___ has joined #pypy

12:56 gef_ has quit [Ping timeout: 256 seconds]

12:56 gef has quit [Ping timeout: 252 seconds]

13:03 <antocuni> cfbolz: where is the paragraph about the new JIT behavior?

13:09 <mattip> antocuni: it is rendered here https://doc.pypy.org/en/latest/release-v7.3.6.html (in the highlights section at the top)

13:09 <cfbolz> mattip: yes, maybe a blog post is the way to go

13:10 <cfbolz> Will try to write it soon

13:10 <antocuni> yes, I also think a blog post would be better. The short paragraph doesn't really explain what's going on and it's hard to understand, IMHO

13:14 Atque has quit [Quit: ...]

13:19 <mattip> cfbolz: is there anything else that needs to go into a release candidate?

13:20 <cfbolz> mattip: you mean in terms of features?

13:20 <cfbolz> Don't think so

13:20 <mattip> cool

14:15 <Dejan> Good stuff

17:45 stkrdknmibalz has joined #pypy

17:58 <the_rat> Dummy question: does PyPy support subinterpreters? I noticed CPython has a plan to have one interpreter lock per interpreter (https://bugs.python.org/issue40512) and currently PyPy uses a global lock defined in C code so I was wondering what kind of work would go into that

17:59 <the_rat> And ultimately whether separate GILs per-interpreter could be implemented in PyPy versions pre-3.10

18:00 <the_rat> (But this breaks C extensions so probably not)

18:03 gef__ has quit [Ping timeout: 252 seconds]

18:03 gef___ has quit [Ping timeout: 256 seconds]

19:04 greedom has joined #pypy

19:14 <mattip> the_rat: it's all very theoretical right now, subinterpreters are not really a thing in CPython yet so it is not clear how they will work

19:14 <mattip> c-extensions are not supported, so that cuts out a lot of the ecosystem (numpy, cython-generated code)

19:45 phlebas has quit [Ping timeout: 276 seconds]

19:50 greedom has quit [Remote host closed the connection]

19:50 greedom has joined #pypy

20:00 phlebas has joined #pypy

20:07 greedom has quit [Quit: Konversation terminated!]

20:07 greedom has joined #pypy

20:14 lritter has joined #pypy

20:16 slav0nic has quit [Ping timeout: 240 seconds]

21:41 greedom has quit [Quit: Konversation terminated!]

21:41 greedom has joined #pypy

21:44 greedom has quit [Client Quit]