#pypy on 2018-01-18 — irc logs at freenode.irclog.whitequark.org

2017-10-17 08:34 antocuni changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | "PyPy: the Gradual Reduction of Magic (tm)"

00:00 dddddd has joined #pypy

00:04 marr has joined #pypy

00:13 antocuni has quit [Ping timeout: 256 seconds]

00:27 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-s390x/builds/829 [default]

00:39 adamholmberg has quit [Remote host closed the connection]

00:39 adamholmberg has joined #pypy

00:39 adamholmberg has quit [Remote host closed the connection]

00:39 adamholmberg has joined #pypy

00:56 Joghurt has joined #pypy

00:56 <Joghurt> Do the recent PyPy releases still support JIT in the ARM builds? The documentation is a little confusing

00:57 <Joghurt> Actually, what I care more about is whether RPython still supports JIT under ARM

00:59 <tumbleweed> yes, they do

00:59 <tumbleweed> only arm7, though

01:00 <tumbleweed> well, and some v6

01:00 <tumbleweed> it needs vfp

01:02 <Joghurt> Good to hear, thanks

01:47 cocoatomo has joined #pypy

01:48 marr has quit [Ping timeout: 276 seconds]

01:48 tbodt has quit [Read error: Connection reset by peer]

01:48 tbodt has joined #pypy

01:55 mvantellingen has quit [*.net *.split]

01:55 pulkitg has quit [*.net *.split]

01:55 LarstiQ has quit [*.net *.split]

01:55 lapinot has quit [*.net *.split]

01:55 MarkMangoba has quit [*.net *.split]

01:55 runciter has quit [*.net *.split]

01:55 idnar has quit [*.net *.split]

01:55 chris| has quit [*.net *.split]

01:55 tito has quit [*.net *.split]

01:55 utek has quit [*.net *.split]

01:55 webmeister has quit [*.net *.split]

01:55 utek has joined #pypy

01:55 LarstiQ has joined #pypy

01:55 tito has joined #pypy

01:55 runciter has joined #pypy

01:55 lapinot has joined #pypy

01:56 MarkMangoba has joined #pypy

01:56 idnar has joined #pypy

01:57 pulkitg has joined #pypy

01:57 webmeister has joined #pypy

01:57 chris|_ has joined #pypy

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/3547 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/5217 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/6505 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/own-linux-x86-32/builds/5681 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/4447 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/3654 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/build-pypy-c-jit-linux-armel/builds/2070 [py3.5]

02:00 <bbot2> Started: http://buildbot.pypy.org/builders/build-pypy-c-jit-linux-armhf-raspbian/builds/1828 [py3.5]

02:05 bjs has quit [Quit: Bye!]

02:07 adamholmberg has quit [Remote host closed the connection]

02:08 adamholmberg has joined #pypy

02:11 tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]

02:12 cocoatomo has quit [Ping timeout: 264 seconds]

02:13 Joghurt has quit [Quit: Leaving]

02:13 adamholmberg has quit [Ping timeout: 240 seconds]

02:40 ceridwen has quit [Ping timeout: 276 seconds]

02:41 mvantellingen has joined #pypy

02:53 ceridwen has joined #pypy

03:01 jcea has quit [Quit: jcea]

03:04 Hotpot33 has quit [Ping timeout: 265 seconds]

03:09 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-64/builds/5217 [py3.5]

03:10 adamholmberg has joined #pypy

03:15 adamholmberg has quit [Ping timeout: 240 seconds]

03:24 ArneBab_ has joined #pypy

03:24 ArneBab_ has quit [Changing host]

03:24 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-64/builds/6505 [py3.5]

03:28 ArneBab has quit [Ping timeout: 256 seconds]

03:29 <bbot2> Failure: http://buildbot.pypy.org/builders/own-linux-x86-32/builds/5681 [py3.5]

03:30 pulkitg has quit [*.net *.split]

03:30 nimaje has quit [*.net *.split]

03:30 kushal has quit [*.net *.split]

03:30 nedbat has quit [*.net *.split]

03:30 bremner has quit [*.net *.split]

03:30 samth has quit [*.net *.split]

03:30 cfbolz has quit [*.net *.split]

03:30 graingert has quit [*.net *.split]

03:30 DRMacIver has quit [*.net *.split]

03:30 __peke__ has quit [*.net *.split]

03:30 nedbat has joined #pypy

03:30 nedbat has quit [Changing host]

03:30 __peke__ has joined #pypy

03:30 nimaje has joined #pypy

03:30 bremner has joined #pypy

03:30 bremner has quit [Changing host]

03:30 samth has joined #pypy

03:31 DRMacIver has joined #pypy

03:31 cfbolz has joined #pypy

03:31 kushal has joined #pypy

03:32 kushal is now known as Guest21826

03:32 graingert has joined #pypy

03:33 pulkitg has joined #pypy

03:33 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-x86-32/builds/4447 [py3.5]

03:33 danieljabailey has quit [Ping timeout: 240 seconds]

03:35 danieljabailey has joined #pypy

03:39 Guest21826 is now known as kushal

03:39 kushal has quit [Changing host]

03:39 kushal has joined #pypy

03:40 marky1991 has quit [Remote host closed the connection]

03:55 <bbot2> Success: http://buildbot.pypy.org/builders/build-pypy-c-jit-linux-armhf-raspbian/builds/1828 [py3.5]

03:55 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-armhf-raspbian/builds/1560 [py3.5]

03:55 <bbot2> Started: http://buildbot.pypy.org/builders/pypy-c-jit-linux-armhf-v7/builds/1399 [py3.5]

03:57 chris|_ is now known as chris|

04:00 <bbot2> Started: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/2443

04:00 <bbot2> Started: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/3243

04:11 <bbot2> Success: http://buildbot.pypy.org/builders/build-pypy-c-jit-linux-armel/builds/2070 [py3.5]

04:12 adamholmberg has joined #pypy

04:17 adamholmberg has quit [Ping timeout: 268 seconds]

04:20 dddddd has quit [Remote host closed the connection]

04:23 kbtr has joined #pypy

04:32 forgottenone has joined #pypy

04:36 drolando has quit [Remote host closed the connection]

04:36 drolando has joined #pypy

04:41 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-win-x86-32/builds/3654 [py3.5]

04:45 <jneen> ooo is set() not supported either then

05:02 raynold has joined #pypy

05:47 Hotpot33 has joined #pypy

05:52 <raynold> ahh it's a wonderful day

06:04 gutworth has joined #pypy

06:20 <kenaan> gutworth default 9a00e4a12145 /: assume math.copysign and always use it over rfloat.copysign

06:21 tayfun26 has joined #pypy

06:31 <arigato> jneen: in rpython? no, no set()

06:35 <gutworth> this @specialize is useless, right? https://bitbucket.org/pypy/pypy/src/9a00e4a12145f5e637ae9890db3de0d44511d32d/pypy/objspace/std/objspace.py?at=default&fileviewer=file-view-default#objspace.py-149

06:35 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-macosx-x86-64/builds/3547 [py3.5]

06:36 <arigato> gutworth: ah, yes

06:37 <kenaan> gutworth default 5e25481ab30d /pypy/objspace/std/objspace.py: no point trying to specialize a NOT_RPYTHON function

06:50 amaury_ has joined #pypy

06:50 realitix has joined #pypy

06:52 forgottenone has quit [Quit: Konversation terminated!]

06:52 amaury_ has quit [Client Quit]

07:05 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-armhf-v7/builds/1399 [py3.5]

07:07 <kenaan> mattip cpyext-datetime2 e3d98c733d8b /: create interp classes to hold the tzinfo field, hack till tests pass and translates

07:51 <fijal> arigato: hi

07:52 <arigato> hi

07:53 <Cheery> hi

07:54 <Cheery> Aside the rlib.rerased, is there anything I could use when implementing a numerics library?

07:55 <Cheery> right now I'll have class Numeric with strategy and storage fields

07:55 <Cheery> and the strategy will also contain reference to 'form', which describes the structure of the numeric.

07:57 <Cheery> I'm looking at it and thinking that it can actually become quite nice when I do little bit of work to it.

07:57 <Cheery> but I might still like to erase one indirection here that appears between the numeric -object and the storage.

07:57 <arigato> we can't give general answers, the question is too vague, but rlib.rerased lets you have several types in a mutable data structure

07:58 <Cheery> arigato: does this help? https://bpaste.net/show/678ed3021ae1

07:59 <arigato> isn't it all immutable?

07:59 <Cheery> it will be yes.

07:59 <arigato> then why do you need rerased? you can just put whatever fields in the subclasses

08:00 <arigato> ah, there are no subclasses of Numeric

08:00 <arigato> so yes, one more standard OO approach would be to make subclasses of Numeric instead of using rerased

08:01 <Cheery> hmmm.. that would perhaps be a good idea.

08:03 <arigato> rerased was really designed as a workaround for an RPython limitation: you can't change the class of an instance

08:03 <Cheery> it would still probably need the mixin -thing.

08:04 <Cheery> I still got the indirection though. I was filling .x, .y, .z etc. with float before this.

08:05 <Cheery> you're right though, this doesn't need the rerased

08:06 <Cheery> it needs something else.

08:09 <arigato> cfbolz: maybe at some point we should fix https://bitbucket.org/pypy/pypy/issues/2728/performance-collapse-with-setattr-in-a ?

08:11 bogner has quit [Remote host closed the connection]

08:11 <cfbolz> arigato: that's 'easy', somebody needs to decide on a threshold when to stop using maps

08:26 Remi_M has joined #pypy

08:45 speeder39 has joined #pypy

08:51 <cfbolz> jneen: you get set performance by using dicts with None values

09:03 kenaan has quit [Read error: Connection reset by peer]

09:13 marr has joined #pypy

09:19 dpn` has quit [Quit: ZNC - http://znc.in]

09:24 antocuni has joined #pypy

09:25 asmeurer_ has quit [Quit: asmeurer_]

09:34 <bbot2> Success: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-32/builds/3243

09:46 bjs has joined #pypy

10:00 dpn` has joined #pypy

10:14 realitix has left #pypy [#pypy]

10:15 Jellyg00se has joined #pypy

10:19 Phi_is_161803398 has joined #pypy

10:19 <Phi_is_161803398> Morning, anyone here?

10:20 <arigato> hi! kind of, yes

10:21 <Phi_is_161803398> Issue I have is as follows: trying to use pypy for a python 3.5-3.6 project. Project uses modules numpy, vegas, and scipy. However, upon trying to install NumPyPy as per the instructions, I get an error saying izip is not in itertools.

10:21 <Phi_is_161803398> My regular python distribution is 3.6 on Anaconda

10:21 <Phi_is_161803398> I made sure to download 10.5.1 with python 3 support from the website

10:22 <arigato> yes? "itertools.izip" doesn't exist in python 3 any more

10:22 <arigato> only in python 2

10:22 <Phi_is_161803398> Yes, so why is PyPy trying to run Python2-only code?

10:23 <arigato> ah, I missed that you try NumPyPy

10:23 <arigato> don't try that one

10:23 <arigato> instead, simply do "pypy3 -m pip install numpy"

10:23 <Phi_is_161803398> Ah. I'll give that whirl

10:23 <Phi_is_161803398> report back in a minute

10:24 <Phi_is_161803398> Yields Microsoft VS errors but I'll manage those.

10:24 <Phi_is_161803398> any plans to make NumPyPy python3 compatible?

10:28 <Phi_is_161803398> It appears pypy3 is still trying to use a python 2 installer

10:30 <Phi_is_161803398> Error message is this: https://bpaste.net/show/782d0bcbc15f

10:31 <Phi_is_161803398> Visual Studio compiler fails with exit code 2

10:31 <arigato> that doesn't tell me anything

10:31 <arigato> paste the whole thing?

10:36 <bbot2> Failure: http://buildbot.pypy.org/builders/pypy-c-jit-linux-armhf-raspbian/builds/1560 [py3.5]

10:38 <Phi_is_161803398> https://bpaste.net/show/a09ef4f5ba8b

10:38 <Phi_is_161803398> 1818 lines :)

10:43 <arigato> sorry, don't know what is wrong (why do you think it's related to Python 2?)

10:44 Grange has joined #pypy

10:45 <Phi_is_161803398> The VS C++ 9.0 is for Python 2.7 specifically

10:45 <Phi_is_161803398> So that's my hunch

10:46 <Phi_is_161803398> I'll see if I can't get a conda env with Python 2.7 up

10:46 <Phi_is_161803398> Thanks for your help :)

10:50 Grange has quit [Ping timeout: 256 seconds]

10:51 <arigato> ah, yes, pypy3 on Windows still tries to call VS C++ 9.0

10:51 Grange has joined #pypy

10:51 antocuni has quit [Ping timeout: 265 seconds]

10:51 <Phi_is_161803398> Good argument for my employer to allow me to use Linux :)

10:52 <Phi_is_161803398> Time for lunch

10:52 Phi_is_161803398 is now known as Phi|Lunch

10:54 speeder39 has quit [Quit: Connection closed for inactivity]

10:58 Grange has quit [Ping timeout: 276 seconds]

11:04 Grange has joined #pypy

11:05 Grange has quit [Client Quit]

11:10 Grange has joined #pypy

11:22 kenaan has joined #pypy

11:22 <kenaan> arigo pypy.org[extradoc] a32c233f73e0 /: Make it clearer that PyPy3 Win32 is beta

11:22 Phi|Lunch is now known as Phi_is_161803398

11:30 <arigato> Phi_is_161803398: been looking more, but it seems that numpy for python 3 fails to compile with this old compiler

11:30 <Phi_is_161803398> Sucky sucky

11:31 <arigato> someone would need to dig inside the generated .c files

11:31 <arigato> alternatively, if you feel like it, you can start digging inside pypy3 to figure out why it still uses that old VC

11:32 <arigato> which doesn't really make sense to me because it is using the new python3 standard library

11:32 <arigato> but maybe it's a part that we modified somehow

11:32 <arigato> (as you can tell I'm no Windows guy)

11:34 <arigato> the final option is that "we can't use beta software" can be used as an argument to switch to linux :-)

11:36 <Phi_is_161803398> Not a Windows guy either

11:36 <arigato> ...ah, no, in order to support a more recent VC we'd need pypy3 itself to be compiled with it too, and that needs tweaks in the buildbots

11:37 <Phi_is_161803398> Great.

11:37 <Phi_is_161803398> I'll try and make do with numba then

11:38 <arigato> sorry about that, we really need someone on windows who wants to drive it forward

11:39 <Phi_is_161803398> Hey, it is what it is

11:39 <Phi_is_161803398> I spent exactly 0 euros on this piece of software

11:40 <Phi_is_161803398> I'm happy that I got help at all

11:40 <Phi_is_161803398> Vegas documentation got taken offline /o\

11:40 <arigato> heh, usually it's the other way around: free software is easier to help with

11:40 <arigato> ...to get help with

11:40 <Phi_is_161803398> Yeah, bigger pool of devs to help

11:41 <Phi_is_161803398> But then it's mostly Linux-based

11:41 <arigato> yesish. Im confident that the Windows part will be fleshed out. it's just that we first released the PyPy3 complete version last month

11:42 <arigato> so it's new

11:42 <Phi_is_161803398> Yeah, it's cool

11:42 <arigato> if python 2 is ok for you, try pypy2 on windows

11:49 yuyichao has quit [Ping timeout: 256 seconds]

11:50 tayfun26 has quit [Quit: tayfun26]

11:50 forgottenone has joined #pypy

11:52 <Phi_is_161803398> I'd have to port all my code to python 2

12:13 marky1991 has joined #pypy

12:13 marky1991 has quit [Remote host closed the connection]

12:14 marky1991 has joined #pypy

12:19 dddddd has joined #pypy

12:25 marky1991 has quit [Ping timeout: 265 seconds]

12:34 antocuni has joined #pypy

12:35 oberstet has joined #pypy

12:46 pilne has quit [Quit: Quitting!]

12:49 larsivi_ has quit [Ping timeout: 256 seconds]

12:55 Phi_is_161803398 has quit []

12:58 marky1991 has joined #pypy

13:11 Rhy0lite has joined #pypy

13:19 Grange has quit [Quit: WeeChat 1.9.1]

13:35 marky1991 has quit [Remote host closed the connection]

13:36 marky1991 has joined #pypy

13:36 amaury_ has joined #pypy

13:55 <bbot2> Success: http://buildbot.pypy.org/builders/jit-benchmark-linux-x86-64/builds/2443

14:13 adamholmberg has joined #pypy

14:27 drolando has quit [Remote host closed the connection]

14:28 drolando has joined #pypy

14:34 amaury_ has quit [Ping timeout: 256 seconds]

14:35 forgottenone has quit [Read error: Connection reset by peer]

14:37 marky1991 has quit [Ping timeout: 256 seconds]

15:03 jcea has joined #pypy

15:03 marky1991 has joined #pypy

15:04 marky1991 has quit [Remote host closed the connection]

15:04 marky1991 has joined #pypy

15:12 <cfbolz> arigato: do you have an opinion from how many attributes on instances should give up and use a dict? 64?

15:21 <arigato> I vaguely thought about a similar number, but maybe we should quickly check, comparing times with and without a call to __dict__={

15:21 <arigato> d.__dict__={}

15:29 cocoatomo has joined #pypy

16:06 forgottenone has joined #pypy

16:10 nunatak has joined #pypy

16:33 yuyichao has joined #pypy

16:37 marky1991 has quit [Ping timeout: 260 seconds]

16:55 jcea has quit [Quit: jcea]

16:58 Jellyg00se has quit [Quit: Leaving]

16:59 jcea has joined #pypy

17:01 marky1991 has joined #pypy

17:02 lasley has quit [Ping timeout: 260 seconds]

17:16 jamesaxl has joined #pypy

17:18 mat^2 has joined #pypy

17:19 marky1991 has quit [Ping timeout: 260 seconds]

17:20 amaury_ has joined #pypy

17:21 amaury_ has quit [Client Quit]

17:21 amaury_ has joined #pypy

17:28 antocuni has quit [Ping timeout: 256 seconds]

17:36 yuyichao_ has joined #pypy

17:36 yuyichao has quit [Read error: Connection reset by peer]

18:00 adamholmberg has quit [Remote host closed the connection]

18:00 asmeurer__ has joined #pypy

18:00 adamholmberg has joined #pypy

18:02 <cfbolz> arigato: eh, indeed. at 64 attributes, mapdict is still faster for a loop that does setattr!

18:02 <cfbolz> but at 80 it isn't

18:02 <cfbolz> I should measure this a bit more systematically

18:02 mattip has joined #pypy

18:05 adamholmberg has quit [Ping timeout: 256 seconds]

18:05 <kenaan> mattip cpyext-datetime2 862b4b1c5eda /pypy/module/cpyext/: make headers compatible with cpython, prevent recursive tp_dealloc

18:09 <mattip> cpyext is really good, I can run untranslated tests to try rpython code that imports functions written in C, a whole module written in pure python,

18:10 <mattip> the classes all inherit from eachother and it just works

18:11 <mattip> in this branch the pure-python module imports interp-level mixed modules, and inherits from the W_* classes

18:16 <mattip> about ensurepip

18:17 adamholmberg has joined #pypy

18:17 <mattip> this page https://pip.pypa.io/en/stable/installing doesn't mention it, are we packaging differently than cpython?

18:34 oberstet has quit [Ping timeout: 240 seconds]

18:46 amaury_ has quit [Ping timeout: 256 seconds]

18:48 juliopy has joined #pypy

18:54 oberstet has joined #pypy

19:04 adamholmberg has quit [Remote host closed the connection]

19:04 adamholmberg has joined #pypy

19:05 adamholmberg has quit [Read error: Connection reset by peer]