arigato changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end ) | use cffi for calling C | mac OS and Fedora are not Windows
k1nd0f_ has joined #pypy
k1nd0f_ has quit [Ping timeout: 246 seconds]
k1nd0f_ has joined #pypy
k1nd0f_ has quit [Ping timeout: 245 seconds]
_whitelogger has joined #pypy
redj has quit [Ping timeout: 245 seconds]
lritter has quit [Ping timeout: 245 seconds]
lritter has joined #pypy
redj has joined #pypy
PileOfDirt has quit [Remote host closed the connection]
redj has quit [Read error: Connection reset by peer]
_whitelogger has joined #pypy
lritter has quit [Remote host closed the connection]
<kenaan> mattip unicode-utf8-py3 27b376c2d467 /pypy/interpreter/test/test_gateway.py: os.fsencode('\x80') fails on python3 windows
dddddd has quit [Remote host closed the connection]
Taggnostr has quit [Ping timeout: 252 seconds]
redj has joined #pypy
_whitelogger has joined #pypy
<cfbolz> mattip: I am working towards a fix of the globals performance problem. Would you be up to merge unicode-utf8 into the py3 version?
<mattip> we lost bencher4, which runs our linux-64 builds. I sent a mail to the maintainers.
<mattip> cfbolz: sure. I did so not so long ago
<mattip> 06c2230fec3c, a week ago
<cfbolz> mattip: yes, but one commit from the meantime I don't understand at all
<mattip> ?
<cfbolz> The one about unicode.title
<cfbolz> Basically py2 and py3 seem to behave differently about this issue
<mattip> interesting
<mattip> I did the py2 fix after looking at the cpython code and the failing test
<cfbolz> Right
<cfbolz> I didn't investigate much, just noticed that py3 behaves differently
<mattip> ok, will look soon
<cfbolz> mattip: thanks
<mattip> indeed, cpython2 and cpython3 behave differently with (unichr(0x345) + u'abc').title()
<mattip> and (chr(0x345) + u'abc').title()
<mattip> so I guess it is ok to merge and ignore those changes
<cfbolz> OK
<cfbolz> Will do that later today
<mattip> thanks
<mattip> any reason to run a benchmark on what you did already in c1ff162569bd or should I wait?
<cfbolz> mattip: yes, it should fix the issue ronan was investigating
<cfbolz> Probably worth it to try his tiny reproducer first
<mattip> ok. need to locally build since the buildbot is down
<cfbolz> :-(
<cfbolz> mattip: the thing I am doing now is more a general improvement for unicode key dicts
<cfbolz> For both 2 & 3
<mattip> great!
<mattip> I spent some time looking at win32, but there are so many failing tsts on py3.5, even without unicode-utf8-py3, I gave up
<mattip> we need some sponsorship for windows
<mattip> solving the pure-cffi missing things in lib_pypy/_winapi.py stuff should not require rpython,
<mattip> and would fix many failing tests
<cfbolz> mattip: yes, agreed with 'funding for Windows'
k1nd0f_ has joined #pypy
<arigato> Me too, fwiw
<arigato> cfbolz: can you quickly describe what you are doing about UnicodeDict?
k1nd0f_ has quit [Remote host closed the connection]
k1nd0f_ has joined #pypy
<mattip> cfbolz: translation is broken, kwargsdict.py has switch_to_unicode_strategy with decode_key_str, which was removed
themsay has joined #pypy
<cfbolz> mattip: ah, damn, will fix
<cfbolz> arigato: yes, so far on the branch unicode dict is ASCII only
<cfbolz> And I am considering to have a unicode dict that stores the wrapped unicode keys, but has a faster hash and equal
<cfbolz> arigato: not completely clear to me whether it should be in addition to the ASCII strategy, or instead. Do you have an opinion?
dddddd has joined #pypy
k1nd0f_ has quit [Ping timeout: 246 seconds]
forgottenone has joined #pypy
<kenaan> cfbolz unicode-utf8-py3 b3a828a76f63 /pypy/objspace/std/test/test_dictmultiobject.py: merge py3.5
<kenaan> cfbolz unicode-utf8-py3 0e7fd7135bc7 /pypy/objspace/std/: disable the fromkeys fast path, it's yet another way to get a non-ascii key
<kenaan> cfbolz unicode-utf8-py3 e75259de9a45 /pypy/objspace/std/: fix the kwargsdict problem
<kenaan> cfbolz unicode-utf8-py3 59c5294dfb36 /pypy/objspace/std/: merge heads
<cfbolz> mattip: should be fixed (at least it annotated for me locally
jcea has quit [Ping timeout: 252 seconds]
waldhar has joined #pypy
<mattip> thanks
jcea has joined #pypy
mattip has quit [Ping timeout: 268 seconds]
mattip has joined #pypy
jcea has quit [Ping timeout: 250 seconds]
jcea has joined #pypy
moei has quit [Ping timeout: 246 seconds]
Zaab1t has joined #pypy
jcea has quit [Ping timeout: 252 seconds]
Ai9zO5AP has quit [Quit: WeeChat 2.3]
k1nd0f_ has joined #pypy
Ai9zO5AP has joined #pypy
k1nd0f_ has quit [Ping timeout: 250 seconds]
waldhar has quit [Quit: ZNC 1.7.1 - https://znc.in]
waldhar has joined #pypy
dmalcolm has quit [Ping timeout: 250 seconds]
Ai9zO5AP has quit [Ping timeout: 244 seconds]
dmalcolm has joined #pypy
Ai9zO5AP has joined #pypy
dmalcolm has quit [Ping timeout: 246 seconds]
dmalcolm has joined #pypy
dmalcolm has quit [Ping timeout: 258 seconds]
jacob22__ has quit [Read error: Connection reset by peer]
dmalcolm has joined #pypy
<kenaan> arigo default d0187cf2f1b7 /pypy/module/pypyjit/test_pypy_c/: Fix the tests for #2904
<arigato> test_xpickle.py I have no clue about
<arigato> this file starts with 15 obscuuuure lines playing around with sys.path and using exec and __name__
<mattip> it uses cpython to unpickle
<mattip> so it looks for python2 or python3 and doesn't find it on the buildbot
<arigato> I think it's more messy than that
<arigato> it fails in the same way on my machine, and I have python2.7
<arigato> also called python2
<mattip> it somehow is not finding the right python to execute
<mattip> or is trying to use our bytecode in that python
<arigato> it used to work, though
<arigato> what changed?
<mattip> hmm
<arigato> ok, no, maybe it never worked. I can't have the test pass on my laptop even using older pypy and corresponding older revision checkouts
<arigato> so maybe it was something like, on the previous buildbot, "python2.7" would actually run pypy
<mattip> passes for me locally with dafacc40cdc4 from August 21
<mattip> I get 506 tests OK
<mattip> ubuntu 18.04
dmalcolm has quit [Ping timeout: 245 seconds]
<mattip> let me see if I can work out what "worker" is calling
dddddd has quit [Ping timeout: 245 seconds]
<mattip> it is calling /usr/bin/python2.7 lib-python/2.7/test/test_xpickle.py worker, and sending it pickled things
<mattip> that is the successful call. Now I need to find a failing one and add the same printing code
zaytsev has joined #pypy
<zaytsev> mattip: hey there
<mattip> hi
<zaytsev> mattip: sorry your email has caught me travelling back to germany from my winter vacation in russia
<zaytsev> mattip: i just did the edit that you suggested, could you please test it or tell me how do i test it?
<mattip> ok
<mattip> go here http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32, see the anubos64 is paused
<zaytsev> mattip: also curious whether you've got my other (way too delayed) answer to the email before that one... you didn't answer, or at least i didn't get your answer, so i assumed the problem is solved... which in fact wasn't the case
<zaytsev> mattip: yes, it is
<mattip> click the anubis64 link, unpause it, click the SalsaSalsa link and pause it
<zaytsev> mattip: roger that
<mattip> then kick a Force build by putting your name in and press "Force Build", and wait
<zaytsev> mattip: shouldn't i supply a branch?
<mattip> "default" is chosen if left empty, which si the main trunk branch
<zaytsev> mattip: yeah, but we want to test pypy3 don't we?
<fijal> zaytsev: winter vacation in russia sounds hardcore ;-)
<mattip> ahh, right. Too many irons in the fire. So put py3.5 in the branch
<bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/4335 [Yury V. Zaytsev: force build, py3.5]
<mattip> +1
<zaytsev> fijal: actually the weather was rather mild this year, like -8 / -12 most of the time, and the lowest was something like -18 while i was there
<fijal> I like cold, but I would not call -12 rather mild
dddddd has joined #pypy
<fijal> but hey, people look at me weird when I go to europe for winter from south africa
themsay has quit [Ping timeout: 272 seconds]
<mattip> "mild" - we get down to +5 and I turn on all the heaters and put on another layer
<zaytsev> fijal: two years ago it went down to -32 and this i must admit is a bit uncomfortable. after 15 min of walking the icicles are hanging down from my beard as the water the the air i breath out condensates and freezes...
<mattip> arigato: actually it all passes for me, even the one from last night. So it is a buildslave issue
<mattip> indeed, linux32 passes that test
<mattip> I added with open('/tmp/log.txt', 'a') as fid: fid.write(sys.executable, sys.argv, '\n') to the worker_main
<mattip> to see what the system call is, if yours is failing maybe that will give us a hint
<zaytsev> fijal: mattip: in fact it's not the temperature itself that is the problem, but rather the fact that winter extends over 6 months. that's no fun. not as bad as in the polar circle, but still. anyways, russia is pretty awesome if you have money (and it can suck pretty badly if you don't). so i really enjoyed it. sports, playing my old piano, no cooking, no work, no colleagues. pipe dream
<mattip> nice
<mattip> zaytsev: we have a sprint in Dusseldorf Feb 4-9
<fijal> zaytsev: where do you live normally if I might ask?
<zaytsev> fijal: sure, i've been living in a small village between aachen and köln (germany) for the last 5 years or so
<zaytsev> fijal: used to work at a nuclear research site they built in a middle of nowhere, so that nobody's sorry if it blows up, and then kind of got stuck there :-/
<fijal> heh
<fijal> I'm right now in Koln
<fijal> just arrived
<zaytsev> mattip: yeah i know, but i'm not sure i will be able to show up even to have a drink. we have to ship our stuff on 01.03 and the hell with break loose if we don't. so no idea if i can make it.
<zaytsev> fijal: i'm going to drive to koln tomorrow, that's where i'm working right now. you know, near this chocolate museum thing...
<fijal> zaytsev: cool, want to have a coffee?
<fijal> me and Armin Rigo are at the IMM
<fijal> till Sunday
<zaytsev> fijal: pardon my ignorance but what's imm
<fijal> an interior design show in koelnmesse
<zaytsev> fijal: sure sounds awesome, i think for me tuesday evening would be best. of course you're welcome to stop by. we have a full-automatic coffee machine and i can buy some finger food. should ask my team if they want to see the legendary pypy core developers irl :) no hologram!
<fijal> cool!
<fijal> tuesday evening works
<zaytsev> fijal: so you are in the business of jit compiling chairs now are you
<fijal> this is our new business
<zaytsev> fijal: wow that sounds brilliant. i'm now in the business of doing boring web stuff for banks and insurance companies. maybe by the end of the year the problem would solve itself though ;-) then i know where i should apply. although i might be too stupid to work with armin. but i'm pretty good at dull ops stuff.
<fijal> zaytsev: we need to make enough money to hire people though first ;-)
<fijal> we are making *some* money, but not quite enough yet
<fijal> but end of the year might look different
<zaytsev> fijal: i'll be entitled to some pretty good unemployment benefits for a year or so, that gives you/us more time ^__^
<fijal> hahaha
<fijal> we sure hope so to make more money
<fijal> that's why we are at the design fair
<fijal> (with the cheapest stand there)
<energizer> fijal: that's so cool!
<energizer> just watched the demo vid, looks awesome
<mattip> for pay
<energizer> mattip: was that for me?
<mattip> energizer: no, irc brain fart, sorry
<energizer> fijal: that's not in python is it?
<fijal> energizer: mostly C#
<energizer> mhmm
<zaytsev> fijal: so how do you define evening? how/where do we meet
<fijal> dinner?
<fijal> I think we have to be at the fair till 6
<fijal> so past that
<zaytsev> fijal: i can pick you up by car and then bring somewhere you need to be. i see there is a lidl near koeln messe, so that sounds like free parking where we can meet
<zaytsev> i checked parking at koeln messe and it's 10 euros. rip off!
<fijal> we also have a car and we can also move by public transit so it's not a problem to meet somewhere
<fijal> so just pick a place
<zaytsev> fijal: we're at im zollhafen, 22 and there is an underground parking lot there. the problem is, it's not free. and i have a permit just for my car :) so you are of course welcome to park there and we can meet at our place
<zaytsev> otherwise as i said i can pick you up and then bring you where you want to be
<zaytsev> fijal: i can show you around and wee can have a dinner at https://www.lecker-essen-suedstadt.de/ - this is where i usually go. normally not much people so don't even need to book a table
<fijal> zaytsev: ok, let's finalize the details tomorrow afternoon?
<zaytsev> fijal: sounds good, i'm gonna post to slack now to see if anyone else is interested and also check if our company wants to sponsor the event :)
<zaytsev> i'll ping you on irc at some point ;-)
<fijal> soudns good
<zaytsev> mattip: so the build is running but i'm afraid it will take awhile. i hopw it will work now, but if it doesn't, then just please drop me an email
Zaab1t has quit [Ping timeout: 268 seconds]
Zaab1t has joined #pypy
Zaab1t has quit [Client Quit]
<mattip> ok
lritter has joined #pypy
k1nd0f_ has joined #pypy
forgottenone has quit [Quit: Konversation terminated!]
k1nd0f_ has quit [Ping timeout: 245 seconds]
<zaytsev> it failed again, but this time with a different problem
<zaytsev> mattip: i think the problem is your magic prebuilt stuff is not built with the same vc version you told me to install
jcea has joined #pypy
jacob22__ has joined #pypy
k1nd0f_ has joined #pypy
dmalcolm has joined #pypy
k1nd0f_ has quit [Ping timeout: 250 seconds]
forgottenone has joined #pypy
k1nd0f_ has joined #pypy
dddddd has quit [Ping timeout: 250 seconds]
dddddd has joined #pypy
moei has joined #pypy
forgottenone has quit [Quit: Konversation terminated!]
k1nd0f_ has quit [Ping timeout: 246 seconds]