cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | "the modern world where network packets and compiler optimizations are effectively hostile"
<njs> fijal: I have some idea what "numpy 2.0 based on memoryviews" refers to, yes
<njs> fijal: I would characterize it as poorly-motivated vaporware right now
Aktariel has joined #pypy
marr has quit [Ping timeout: 260 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 260 seconds]
tbodt has joined #pypy
ArneBab_ has joined #pypy
ArneBab has quit [Ping timeout: 248 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 248 seconds]
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
pilne has joined #pypy
jcea has quit [Ping timeout: 246 seconds]
rokujyouhitoma has joined #pypy
Niedar has quit [Ping timeout: 255 seconds]
wallet42 has quit [Ping timeout: 255 seconds]
sbauman has quit [Remote host closed the connection]
altendky has quit [Remote host closed the connection]
MarkMangoba has quit [Remote host closed the connection]
rokujyouhitoma has quit [Ping timeout: 268 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
pilne has quit [Quit: Quitting!]
<fijal> njs: I agree with "if it sounds simlpe, maybe someone didn't think about it hard enough"
_whitelogger has joined #pypy
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 260 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 255 seconds]
oberstet has joined #pypy
_whitelogger has joined #pypy
rokujyouhitoma has joined #pypy
_main_ has joined #pypy
_main_ has quit [Read error: Connection reset by peer]
_main_ has joined #pypy
__main__ has quit [Read error: Connection reset by peer]
rokujyouhitoma has quit [Ping timeout: 246 seconds]
__main__ has joined #pypy
_main_ has quit [Read error: Connection reset by peer]
vkirilichev has joined #pypy
rmariano has joined #pypy
ronan has joined #pypy
antocuni has joined #pypy
arigato has joined #pypy
marr has joined #pypy
rokujyouhitoma has joined #pypy
marr has quit [Remote host closed the connection]
rokujyouhitoma has quit [Ping timeout: 248 seconds]
kenaan has joined #pypy
<kenaan> arigo py3.6 68e743f6f11e /pypy/interpreter/: (vxgmichel, arigo) Handle exhausted async generators correctly
rmariano has quit [Ping timeout: 260 seconds]
<nimaje> wait, should that be supported? but it seems to be right as it is like ´for x in range(10): yield (yield x)´
<fijal> nimaje: I don't think it should be supported tbh
<nimaje> in pypy3 it gives a SyntaxWarning linking that issue http://bugs.python.org/issue10544 :)
<fijal> hehehe
<nimaje> well, it works in cpython 2.7.13, 3.6.1, pypy and pypy3 5.8.0 (I only tested these)
<fijal> "works"
<kenaan> arigo py3.6 2d1ce02f13b7 /pypy/interpreter/: in-progress
<nimaje> as I said I would expect that result, but I didn't expect it to be valid :)
<fijal> nimaje: I didn't
rokujyouhitoma has joined #pypy
<nimaje> fijal: what did you expect? try send on that generator :)
<fijal> I don't know
rokujyouhitoma has quit [Ping timeout: 246 seconds]
<kenaan> arigo py3.6 8a655968dc6f /pypy/interpreter/: (vxgmichel, rmariano, arigo) Start on async generators that contain "await"
nimaje1 has joined #pypy
nimaje1 is now known as nimaje
nimaje is now known as Guest85147
Tiberium has joined #pypy
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 248 seconds]
<kenaan> arigo py3.6 bc59a0b5e616 /pypy/: (vxgmichel, arigo) asend/send on async_generator objects
cstratak has joined #pypy
<kenaan> arigo py3.6 4d7a6b1e53da /pypy/interpreter/test/test_coroutine.py: More tests, not passing
rmariano has joined #pypy
vkirilichev has quit [Remote host closed the connection]
<kenaan> arigo default 0b72fd1a7641 /pypy/interpreter/test/test_generator.py: Add a test that is passing on pypy2.7, cpython2.7, cpython3.x, but not pypy3.5
<kenaan> arigo py3.5 b58d747dcf74 /: hg merge default
<kenaan> arigo py3.6 d3b14669d2ed /pypy/interpreter/generator.py: fix for test_async_yield_explicit_asend_used_several_times
vkirilichev has joined #pypy
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
<kenaan> arigo py3.5 f6faa2c5c765 /pypy/module/_vmprof/interp_vmprof.py: rename args
pilne has joined #pypy
antocuni has quit [Ping timeout: 240 seconds]
<kenaan> rmariano py3.5 0e1f2cd91ced /pypy/interpreter/generator.py: (arigo, rmariano) Fixing regression with non-started generator receiving non-None, should always raise TypeError.
<kenaan> rmariano py3.5 6de61c965087 /pypy/module/_vmprof/interp_vmprof.py: Merge heads
bauerj has quit [Quit: No Ping reply in 180 seconds.]
<kenaan> rmariano py3.6 51ae4831bf9b /: Merge branch py3.5
bauerj has joined #pypy
<kenaan> p_zi...@yahoo.de py3.6 caa89dbcd27a /pypy/: (ronan,pzieschang) fixed int() behaviour to mirror python 3.6, where a real int must be returned
<kenaan> rlamy py3.6 f593f803fcdb /pypy/: Merged in mad-marty/pypy/py3.6 (pull request #556) (ronan,pzieschang) fixed int() behaviour to mirror python 3.6, wh...
marr has joined #pypy
<kenaan> rmariano py3.6 6f9eb0f25f77 /pypy/interpreter/test/test_coroutine.py: Add test for async generators with asend and throw, and vice-versa
<kenaan> rmariano py3.6 1739c19d3d16 /pypy/: Merge heads
<kenaan> arigo py3.6 75e725412a83 /pypy/interpreter/generator.py: AsyncGenASend().descr_throw()
cstratak has quit [Quit: Leaving]
rokujyouhitoma has joined #pypy
<kenaan> p_zi...@yahoo.de py3.6 3ed5101a25c4 /pypy/objspace/std/: (ronan,pzieschang) fixed int() behaviour also for __trunc__ returning a subclass of int
<kenaan> rlamy py3.6 b69f41d7b76d /pypy/objspace/std/: Merged in mad-marty/pypy/py3.6 (pull request #557) (ronan,pzieschang) fixed int() behaviour also for __trunc__ retur...
rokujyouhitoma has quit [Ping timeout: 248 seconds]
altendky has joined #pypy
<kenaan> arigo py3.6 32f19625dc25 /pypy/interpreter/: support for athrow()
<kenaan> fijal extradoc a123560be7b9 /talk/pyconru-2017/talk.key: add my talk draft
vkirilichev has quit [Remote host closed the connection]
antocuni has joined #pypy
jcea has joined #pypy
redj has quit [Quit: No Ping reply in 180 seconds.]
redj has joined #pypy
lritter has joined #pypy
<kenaan> p_zi...@yahoo.de py3.6 0c4aaad09618 /pypy/objspace/std/test/test_intobject.py: (ronan,pzieschang) fixed test_intobject behaviour for python 3.6 as well with bool
<kenaan> rlamy py3.6 6e9e65d8b76a /pypy/objspace/std/test/test_intobject.py: Merged in mad-marty/pypy/py3.6 (pull request #558) (ronan,pzieschang) fixed test_intobject behaviour for python 3.6 ...
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
<kenaan> rmariano py3.6 1bf8606b71de /pypy/interpreter/test/test_coroutine.py: Add test for CPython for atrow().throw In this case CPython raises RuntimeError, similar as if it was a wrong par...
<antocuni> arigato: I think that commit 05b724e broke ABI compatibility
<antocuni> because if I try to load a .so module compiled with an older pypy version, I get "undefined symbol: PyMem_Malloc"
<antocuni> I was about to just update the DEFAULT_SOABI version in module/imp/importing.py
<antocuni> however, the comment above is a bit scary :)
asmeurer__ has joined #pypy
<kenaan> rlamy py3.5 fe19ba9deaa2 /pypy/objspace/std/bytesobject.py: fix for 6159e89116af
<kenaan> rlamy py3.5 75b45f45de1b /pypy/objspace/std/bytesobject.py: Improve performance of bytearray.extend(<bytes>)
<kenaan> arigo py3.6 4146af845534 /pypy/interpreter/: Relax and implement the latest test
<arigato> antocuni: yes, I'm not sure what to do
<antocuni> I don't understand why we can't simply update it
<arigato> it would break all cffi-built .so, too, for no reason
<antocuni> is this the only reason? I.e., newer cffi's module will get a .pypy-59.so extension
<antocuni> but if you recompile, stuff will simply work?
<arigato> yes, and already-compiled modules with .pypy-41.so won't load, although recompiling would produce the very same file
<antocuni> I see
<arigato> I would be ok if we break this compatibility once and call the cffi modules ".pypy-cffi.so" or something
<arigato> then we don't have the issue any more
<antocuni> I don't think it's a big problem, honestly; people kind of expect they need to recompile things if you update pypy
<antocuni> plus, the vast majority of people uses some kind of automated build step which involves virtualenv and/or docker or similar tools
<arigato> I know that I can hack my virtualenvs by changing which version of pypy runs with it
<arigato> and it works
<fijal> anyone feel like reviewing the slides for tomorrow?
<antocuni> yeah, I agree, until it stops working because cpyext is broken
* antocuni reviewing the slides
<arigato> what kind of projector can display 5 big words but not graphs? :-)
<antocuni> I was about to ask the same
<simpson> fijal: Awesome slides. I hope to share the recorded talk with folks at some point, because this is awesome.
<fijal> arigato: 800x600 VGA
<fijal> arigato: the graph in question is a complicated matrix from here
<antocuni> fijal: I think that the age of "solved problems" is closer to 30 than 20 years
<fijal> antocuni: fine ;-)
<antocuni> fijal: also, about the slide "culture": if you need an example of project which was written in pure-python with performance in mind from the beginning, you can mention capnpy
<antocuni> it's impressively fast :)
<fijal> cool will do
<simpson> Self moved to Sun in 1991, so yeah, closer to 30 than 20 at this point. How many more decades will it be...?
<antocuni> apart from that, good slides
<arigato> yes, good slides
asmeurer__ has quit [Quit: asmeurer__]
asmeurer_ has joined #pypy
asmeurer_ has quit [Client Quit]
<LarstiQ> fijal: "the python as a language" is maybe more idiomatic English with s/the//
<fijal> LarstiQ: ok
<fijal> LarstiQ: I'm speaking to russians btw
<fijal> so for them "the" is as alien as for me
<LarstiQ> :)
tormoz has joined #pypy
<LarstiQ> fijal: explains the cyrillic, looks like a good talk
<LarstiQ> how is this related to sourced.tech?
<fijal> not at all?
<fijal> sourced.tech seems like serious research
<fijal> all I'm proposing is serious engineering :)
<LarstiQ> k, my vague impression is that sourced.tech has some Russian connection, and there was this one slide
<rmariano> Hi
<arigato> hi co-sprinter
<kenaan> rmariano py3.6 b6e947786fa5 /pypy/interpreter/test/test_coroutine.py: Add test for async generator fail while running Check when an exception is thrown by calling `.athrow()` while on...
rmariano has quit [Quit: Page closed]
rokujyouhitoma has joined #pypy
ronan has quit [Ping timeout: 240 seconds]
Tiberium has quit [Remote host closed the connection]
rokujyouhitoma has quit [Ping timeout: 268 seconds]
arigato has quit [Ping timeout: 268 seconds]
asmeurer has joined #pypy
<fijal> antocuni: I found smoeone who read your PhD
<fijal> template <typename idx>
<fijal> __m256i idxvec = _mm256_setr_epi32(0, 1, 2, 3, 4, 5, 6, 7);
<fijal> idx dim, idx i, const float *assigncost, const float *v) {
<fijal> std::tuple<float, float, idx, idx> find_umins(
<fijal> __attribute__((always_inline)) inline
<fijal> __m256 uminvec = _mm256_set1_ps(std::numeric_limits<float>::max()),
<fijal> __m256i j1vec = _mm256_set1_epi32(-1), j2vec = _mm256_set1_epi32(-1);
<fijal> usubminvec = _mm256_set1_ps(std::numeric_limits<float>::max());
<fijal> for (idx j = 0; j < dim - 7; j += 8) {
<fijal> __m256 acvec = _mm256_loadu_ps(assigncost + i * dim + j);
<fijal> __m256 vvec = _mm256_loadu_ps(v + j);
<fijal> __m256 h = _mm256_sub_ps(acvec, vvec);
<fijal> __m256 cmp = _mm256_cmp_ps(h, uminvec, _CMP_LE_OQ);
<fijal> usubminvec = _mm256_blendv_ps(usubminvec, uminvec, cmp);
<fijal> LarstiQ: scary code
<fijal> also, as usual with VC funded companies, there is like 500 people
[0__0] has quit [Ping timeout: 268 seconds]
<LarstiQ> fijal: yeah
<LarstiQ> fijal: wonder how well it works in pypy
vkirilichev has joined #pypy
<fijal> LarstiQ: which part?
<LarstiQ> pyLAPJV/https://github.com/src-d/lapjv/a native python implementation
<LarstiQ> eh, using / as a separator with then a url was a bad idea
<LarstiQ> fijal: mostly that src-d implementation
<antocuni> fijal: cool, say "congrats" to him :)
<fijal> I invited him to a sprint, if we make one closer than CT
<antocuni> it sounds like a safe assumption
<kenaan> antocuni cpyext-jit 1327063282d7 /: experimental branch in which to make cpyext visible to the JIT
<kenaan> antocuni cpyext-jit ae94f6ae3278 /pypy/module/pypyjit/policy.py: remove cpyext from the JIT policy blacklist
<kenaan> antocuni cpyext-jit aee286949d64 /pypy/module/: add enough @jit.dont_look_inside so that pypy translates
<kenaan> antocuni cpyext-jit 25ba96ce5970 /pypy/module/cpyext/: merge part of the cpyext-callopt branch, up to commit 9cbc8bd76297: this should simplify and speedup the call...
<antocuni> apparently, I made cpyext ~6.5 times faster
<antocuni> (running the very simple benchmark here: https://github.com/antocuni/cpyext-benchmarks)
rokujyouhitoma has joined #pypy
<fijal> sick?
<fijal> how much bigger is the binary?
<antocuni> 83MB vs 81MB
<antocuni> but I think the biggest speedup comes to the callopt optimization which we did in leysin back in january
<antocuni> s/comes to/comes from
asmeurer has quit [Quit: asmeurer]
rokujyouhitoma has quit [Ping timeout: 260 seconds]
asmeurer has joined #pypy
<antocuni> fijal: for completeness: pypy 5.8 takes 1.7 secs, my branch 0.26, cpython 0.72
<fijal> cool
<fijal> what happens if you pass args?
<fijal> can we declare that we're officially faster than cpython on Cpyext?
<antocuni> nope
<fijal> antocuni: I prpmised we would write blog post about gil
<fijal> damn
<antocuni> anything useful is still slow
<antocuni> like, if you try to read items from a numpy array
<antocuni> but the cool news is that we probably have a lot of low hanging fruits to take
<fijal> right
<fijal> I would start with passing ints?
<antocuni> fijal: ok, so METH_O are still fast
<antocuni> METH_VARARGS are horribly slow
<antocuni> like, 4.82 seconds instead of 0.3
<fijal> whack whack whack...
<antocuni> (on cpython, METH_VARARGS takes 1.15 vs 0.8 of METH_NOARGS or METH_O)
<antocuni> anyway, /me --> off
mattip has joined #pypy
antocuni has quit [Ping timeout: 255 seconds]
<Cheery> libuv vendors have removed the support for the visual studio 2009 compiler
rokujyouhitoma has joined #pypy
vkirilichev has quit [Remote host closed the connection]
vkirilichev has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
<LarstiQ> https://bitbucket.org/pypy/pypy/issues/2579/pypy35-580-memory-leak seems closeable (can't see a way in the ui for me to do that)
asmeurer has quit [Quit: asmeurer]
deep-book-gk_ has joined #pypy
deep-book-gk_ has left #pypy [#pypy]
tote has quit [Remote host closed the connection]
[0__0] has joined #pypy
oberstet has quit [Ping timeout: 248 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 260 seconds]
asmeurer__ has joined #pypy
asmeurer__ has quit [Ping timeout: 240 seconds]
tilgovi has joined #pypy
asmeurer_ has joined #pypy
<mattip> LarstiQ: done, thanks
tilgovi has quit [Ping timeout: 276 seconds]
rokujyouhitoma has joined #pypy
tilgovi has joined #pypy
yuyichao_ has quit [Read error: Connection reset by peer]
yuyichao has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
antocuni has joined #pypy
asmeurer_ has quit [Quit: asmeurer_]
oberstet has joined #pypy
<LarstiQ> mattip: thanks :)
asmeurer__ has joined #pypy
rokujyouhitoma has joined #pypy
TheAdversary has quit [Remote host closed the connection]
TheAdversary has joined #pypy
rokujyouhitoma has quit [Ping timeout: 260 seconds]
kolko has quit [Quit: ZNC - http://znc.in]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 240 seconds]
asmeurer__ has quit [Quit: asmeurer__]
tbodt has joined #pypy
tilgovi has quit [Ping timeout: 240 seconds]
antocuni has quit [Ping timeout: 255 seconds]
rokujyouhitoma has joined #pypy
oberstet has quit [Ping timeout: 260 seconds]
asmeurer__ has joined #pypy
tilgovi has joined #pypy
lritter has quit [Remote host closed the connection]
rokujyouhitoma has quit [Ping timeout: 248 seconds]
altendky has quit [Remote host closed the connection]
mattip has left #pypy ["bye"]
sbauman has joined #pypy
wallet42 has joined #pypy
MarkMangoba has joined #pypy
altendky has joined #pypy
Niedar has joined #pypy
kipras is now known as kipras`away
raynold has joined #pypy
vkirilichev has quit [Remote host closed the connection]
rokujyouhitoma has joined #pypy
asmeurer__ has quit [Quit: asmeurer__]
rokujyouhitoma has quit [Ping timeout: 260 seconds]
<Cheery> How would rpython compile with http://landinghub.visualstudio.com/visual-cpp-build-tools ?
<simpson> No idea, but it sounds like an adventure.
<Cheery> At least I have managed to compile the dependencies on win32 myself now.
<Cheery> It just feels like a mess though.