cfbolz changed the topic of #pypy to: #pypy PyPy, the flexible snake https://pypy.org | IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end and https://libera.irclog.whitequark.org/pypy | insert pithy quote here
fotis has joined #pypy
Gustavo6046_ has joined #pypy
Gustavo6046 has quit [Ping timeout: 268 seconds]
Gustavo6046_ is now known as Gustavo6046
fotis has quit [Ping timeout: 244 seconds]
fotis has joined #pypy
fotis has quit [Ping timeout: 244 seconds]
fotis has joined #pypy
fotis has quit [Ping timeout: 244 seconds]
fotis has joined #pypy
fotis has quit [Ping timeout: 272 seconds]
fotis has joined #pypy
Gustavo6046 has quit [Ping timeout: 252 seconds]
Gustavo6046 has joined #pypy
fotis has quit [Ping timeout: 244 seconds]
<cfbolz> mattip: nice, general progress
fotis has joined #pypy
fotis has quit [Ping timeout: 272 seconds]
fotis has joined #pypy
<mattip> cfbolz: what's next?
fotis has quit [Ping timeout: 268 seconds]
<cfbolz> mattip: I don't actually know 😅
<cfbolz> We should maybe release an alpha at the end of summer?
fotis has joined #pypy
fotis has quit [Ping timeout: 252 seconds]
<cfbolz> mattip: I suppose I should fix the setting of the current lineno on frames at some point
<cfbolz> that's the next big source of several failures
<cfbolz> ah, and strftime has a missing feature too
fotis has joined #pypy
fotis has quit [Ping timeout: 268 seconds]
fotis has joined #pypy
fotis has quit [Ping timeout: 244 seconds]
<cfbolz> ah no, strftime seems shallow
fotis has joined #pypy
fotis has quit [Ping timeout: 268 seconds]
<mattip> it would be nice to cleanup the own test failures
<mattip> any idea what is up with zipimport?
fotis has joined #pypy
<mattip> something is failing vectorcall tests in lib-python3/test/test_call.py
<mattip> lib-python3/test/test_type_comments.py has errors
otisolsen70 has joined #pypy
<cfbolz> mattip: yeah, many details left
<cfbolz> agreed about our own tests, that's really annoying
fotis has quit [Ping timeout: 240 seconds]
<bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7735 [Carl Friedrich Bolz-Tereick: force build, py3.8-ast-end-positions]
<mattip> we need a tk expert to figure out the 32 failures in test_idle, something about passing arguments
<cfbolz> mattip: yeah, there's also a tk problem around unicode somehow
<mattip> fails the same on py3.7
<cfbolz> yeah
<mattip> maybe the pypy-dev mailing list?
<cfbolz> I looked for 15 min at some point, not obvious
<cfbolz> mattip: worth a try
<cfbolz> mattip: it's a bit unclear why the datetime problems don't appear on cpython, don't they test the pure python implementations?
<mattip> :shrug:
<cfbolz> Same for the asyncio 'accelerators'
fotis has joined #pypy
<cfbolz> Wow, the cpython 3.11 refactorings are really extensive
<fijal> what are they doing?
<cfbolz> fijal: the frame objects don't have their own array anywmore
<cfbolz> instead it's a thread local single array of all the locals+cells+stack
<bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/7735 [Carl Friedrich Bolz-Tereick: force build, py3.8-ast-end-positions]
<cfbolz> fijal: except for generators, of course
<cfbolz> it means they immediately need to think about frames escaping if somebody stores a reference to the result of sys._getframe
<cfbolz> I wonder whether they look at the refcount for that?
fotis has quit [Ping timeout: 244 seconds]
fotis has joined #pypy
<cfbolz> mattip: cpython killed the test about strftime(format=...) instead: https://github.com/python/cpython/blob/3.8/Lib/test/datetimetester.py#L1449
fotis has quit [Ping timeout: 240 seconds]
<fijal> uh
<fijal> ok?
<fijal> what's the point of that? speed?
otisolsen70 has quit [Quit: Leaving]
<mattip> cfbolz: so it was a bad merge?
<mattip> no, it is pypy-specific code added in 4e110113b03d to ensure the pure-python implementation, which differs from cpython, is correct
<mattip> as for zipimport, at least one difference has to do with the case where both a pyc and py file are in the zipimport, and they differ
<mattip> python3.8 will prefer the py, python3.7 will prefer the pyc
<mattip> if I remove the py and leave only the pyc, then pypy3.8 will load the pyc
<mattip> or maybe like usual I am confused
<cfbolz> mattip: I wonder why it does not just work, import is pure python
Gustavo6046 has quit [Quit: ZNC 1.8.2 - https://znc.in]
Gustavo6046 has joined #pypy
<mattip> right, so can we just kill the own zipimport tests?
<mattip> maybe they made sense on python2
<cfbolz> mattip: ah, it's our own tests that fail, not CPython's?
<mattip> yup
<cfbolz> I thought some lib-python failures too?
<cfbolz> If not, then we could remove them (or, if we want to be careful, check whether they work on cpy3.8)
<mattip> they don't work on cpy3.8, and they take 534 secs on the buildbot
<mattip> x86_64, arm is probably much slower
<mattip> they fail on cpy3.8 like they do untranslated
<mattip> no, sorry, they fail differently
fotis has joined #pypy
<cfbolz> mattip: killing seems fine
fotis has quit [Ping timeout: 268 seconds]
<mattip> ahh, hang on, we have a rpython zipimport
<cfbolz> Ah
<cfbolz> So it's likely a real problem
<mattip> the pure-python one in cpython/Lib/zipimport.py is new for cpython 3.8
<mattip> it seems
<mattip> so maybe we should just disable ours for py3.8
<cfbolz> mattip: ah, good
<mattip> cool. So skip the tests and disable the config option
<mattip> actually, might as well kill it
<bbot2> Started: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/8743 [mattip: force build, py3.8]
the_rat has joined #pypy
stkrdknmibalz has quit [Quit: WeeChat 3.0.1]
<cfbolz> mattip: nice
<cfbolz> So all in all the whole importlib move is really great for pypy
stkrdknmibalz has joined #pypy
idnar has joined #pypy
idnar has quit [Changing host]
<bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/8743 [mattip: force build, py3.8]
fotis has joined #pypy
fotis has quit [Ping timeout: 268 seconds]
fotis has joined #pypy
fotis has quit [Ping timeout: 244 seconds]
fotis has joined #pypy
fotis has quit [Ping timeout: 268 seconds]