cfbolz changed the topic of #pypy to: #pypy PyPy, the flexible snake | IRC logs: and | so many corner cases, so little time
lritter has quit [Quit: Leaving]
Guest3959 has quit [Quit: QUIT]
epony has joined #pypy
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
samth has quit [*.net *.split]
[Arfrever] has quit [*.net *.split]
dmalcolm__ has quit [*.net *.split]
infernix has quit [*.net *.split]
mattip has quit [*.net *.split]
lazka has quit [*.net *.split]
dustinm has quit [*.net *.split]
[Arfrever] has joined #pypy
dustinm has joined #pypy
dmalcolm__ has joined #pypy
mattip has joined #pypy
lazka has joined #pypy
infernix has joined #pypy
samth has joined #pypy
Guest96 has joined #pypy
<ctismer> mattip: Are your fixes in the nightly builds, already?
Guest96 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest96 has joined #pypy
<mattip> ctismer: the separation of python-method and c-api-method is in, yes
<ctismer> mattip: Great, then I can remove some hacks :)
Techcable has quit [Remote host closed the connection]
Techcable has joined #pypy
Guest96 has quit [Quit: Textual IRC Client:]
otisolsen70 has joined #pypy
<cfbolz> mattip: cool
<cfbolz> mattip: did you look into the details of that yet? looks reasonable?
<mattip> it seems that only relating to the data apis will simplify the hpy port of numpy
<mattip> since it does not have structured nor object dtypes
<mattip> which makes deallocating much easier
<cfbolz> mattip: right
<cfbolz> mattip: how big is the API? eg would an rpython port be more reasonable again?
<cfbolz> mattip: did you notice that test_threading now passes on 3.9? :-)
<mattip> cfbolz: it might make sense to revive micronumpy
<mattip> test_threading passes, but I think the compileall hang is still there
slav0nic has joined #pypy
<ctismer> mattip: ob_base is no longer in PyVarObject ?
<ctismer> (can be something else, I just suddenly can't build)
<mattip> weird
<mattip> has a py3.9 result, it seems to be the same as py3.7
<mattip> ctismer: last nights nightlies are working with numpy
<ctismer> mattip: ok fine, again 3.8 or is 3.9 no longer alpha?
<mattip> 3.8rc1 is "final", 3.9rc1 is "beta"
<ctismer> mattip: thanks.
<cfbolz> mattip: yeah, I’ve given up for now on the deadlock after a week of staring
<mattip> did you get anywhere figuring out what is happening (even without solving it)?
<cfbolz> There are several different failure modes that I see
<cfbolz> Sometimes a forked process doesn’t even start communicating
<cfbolz> And sometimes the result of one of the tasks gets lost
<cfbolz> But nothing really felt like a pypy bug
<mattip> hmm. I guess I didn;t really check that after the fixes to test_threading the compileall hang still happens
<mattip> fixes to _posixsubprocess that is
<cfbolz> I maintain that they are unrelated
<mattip> yeah, 99% likely
<cfbolz> mattip: anyway it’s very cool that we have posixsubprocess
<cfbolz> In other news , I fixed another roadblock the new dictionary update optimisation, I need to check whether that’s enough to fix the 3.9 problems
greedom has joined #pypy
Atque has quit [Ping timeout: 276 seconds]
Atque has joined #pypy
greedom has quit []
<ctismer> mattip: cfbolz: But in the latest alpha that crashes our build.
otisolsen70_ has joined #pypy
otisolsen70 has quit [Ping timeout: 240 seconds]
<cfbolz> ctismer: hm, something looks wrong here, but I think that's a question for mattip
danchr_ is now known as danchr
iwkse has quit [Ping timeout: 245 seconds]
otisolsen70_ has quit [Quit: Leaving]
iwkse has joined #pypy
<mattip> ctismer: that is coming from setuptools.
<mattip> we bundle setuptools-56.0.0-py3-none-any.whl with py3.8, which has that bug
<mattip> we bundle setuptools-58.1.0-py3-none-any.whl with py3.9 which does not have that bug
<mattip> so we should update the bundled setuptools for the release. Thanks for finding this
<mattip> done
slav0nic has quit [Ping timeout: 250 seconds]