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"
rokujyouhitoma has quit [Ping timeout: 260 seconds]
<kenaan>
wlav cppyy-packaging a0f6d0e8e563 /pypy/module/_cppyy/: rename destruct -> __destruct__ to not interfere with user defined methods
<kenaan>
wlav cppyy-packaging 80bd00a75270 /pypy/module/_cppyy/: remove backend code (lives in http://bitbucket/wlav/cppyy-backend and is shared with CPython) and the builti...
<kenaan>
wlav cppyy-packaging f141497a5252 /pypy/module/_cppyy/genreflex-methptrgetter.patch: remove old genreflex method ptr patch (not needed with Cling)
rokujyouhitoma has quit [Ping timeout: 240 seconds]
oberstet has joined #pypy
vkirilichev has joined #pypy
vkirilichev has quit [Remote host closed the connection]
vkirilichev has joined #pypy
bremner has quit [Ping timeout: 246 seconds]
bremner has joined #pypy
bremner has quit [Changing host]
bremner has joined #pypy
rokujyouhitoma has joined #pypy
forgottenone has quit [Quit: Konversation terminated!]
rokujyouhitoma has quit [Ping timeout: 246 seconds]
antocuni has joined #pypy
exarkun has quit [Ping timeout: 246 seconds]
forgottenone has joined #pypy
exarkun has joined #pypy
Tiberium has joined #pypy
forgottenone has quit [Quit: Konversation terminated!]
cstratak has joined #pypy
oberstet has quit [Ping timeout: 240 seconds]
forgottenone has joined #pypy
yuyichao_ has quit [Remote host closed the connection]
marky1991 has quit [Ping timeout: 276 seconds]
oberstet has joined #pypy
rokujyouhitoma has joined #pypy
<kenaan>
arigo default c8577e0e0932 /rpython/rlib/_rsocket_rffi.py: Issue #2607 (jean-louis) Instead of sys/poll.h, use poll.h, which is the officially correct header.
rokujyouhitoma has quit [Ping timeout: 260 seconds]
Tiberium has quit [Remote host closed the connection]
Tiberium has joined #pypy
Tiberium has quit [Ping timeout: 240 seconds]
rokujyouhitoma has joined #pypy
rokujyouhitoma has quit [Ping timeout: 260 seconds]
antocuni has quit [Ping timeout: 258 seconds]
marr has joined #pypy
jcea has joined #pypy
<kenaan>
arigo default c88684d6f4c1 /pypy/module/_vmprof/test/test__vmprof.py: Skip unreliable test when testing with either -A or with a pypy host
<kenaan>
arigo default 8a0f195540eb /pypy/doc/whatsnew-head.rst: Fix the description (there was never a --nopax option; there was only one temporarily inside that branch)
nimaje1 has joined #pypy
nimaje1 is now known as nimaje
nimaje has quit [Killed (cherryh.freenode.net (Nickname regained by services))]
lritter has joined #pypy
antocuni has joined #pypy
Fleche has joined #pypy
toolchains_x86 has joined #pypy
<toolchains_x86>
Hello, I am trying to compiling PyPy from source, and it seems to take too long even on a machine with a large RAM. So, I wanted to ask why the -jN option is not useable? Thanks!
<ndash>
toolchains_x86: the pypy translation process can't be parallelized
<fijal>
ndash: it can be, it just isn't :)
<ndash>
fijal: details!
<ndash>
toolchains_x86: when the C it generates is compiled, it does use multiple cores
<exarkun>
How does one run a single test case from the lib-python suite?
<fijal>
I don't know if you can
<fijal>
you can run a file though?
<fijal>
exarkun: what I usually do is to start with porting the offending one somewhere saner
<fijal>
where I can run them one by one
<exarkun>
yea, I see how to run a file. some of these files take, like, an hour to run :/
<fijal>
(and not invoke the whole fucking unittest thing)
<exarkun>
I forgot how primitive the stdlib test suite is
mattip has joined #pypy
<exarkun>
okay.
<fijal>
yeah it's terrible
<toolchains_x86>
@ndash when I'm compiling only one core shows 100% usage.
<toolchains_x86>
In a four core machine.
<exarkun>
toolchains_x86: On my 4 core machine, when the translation gets to the C compilation phase, it runs `make -j3 ...`
<exarkun>
with the normal/recommended translation command: `../../rpython/bin/rpython --opt=jit`
<fijal>
toolchains_x86: yes, it runs on one core for the vast majority of the time (as previously mentioned it's not parallelized)
yuyichao_ has joined #pypy
<antocuni>
toolchains_x86: if you have MAKEFLAGS, then this takes the precedence and rpython won't add any '-j' option
<antocuni>
if you don't have MAKEFLAGS, it should automatically add a -j which depends on the number of available cores
<antocuni>
look at rpython/config/support.py
<exarkun>
but the long and short of it really is that, yes, translation takes a while.
<exarkun>
fijal: Can I get a recently translated pypy3 from pypy buildbot or someplace like that?
<toolchains_x86>
Perhaps I don't understand the whole process quite clearly as at now.. How do I inject the translated binary into the cross-compile envronment, for instance
marky1991 has joined #pypy
<LarstiQ>
toolchains_x86: you're cross compiling because openwrt's targets have severely limited cpu/memory?
<exarkun>
mattip: more visible links to it might be a start. I spent some time on http://doc.pypy.org/en/latest/build.html which just has a brief section on hg. maybe a link to the getting started guide from the build page would be helpful.
<antocuni>
IIRC at some point we did support cross compilation for nokia maemo
<antocuni>
but I'm not really sure what is the current status
yuvipanda has quit [Ping timeout: 246 seconds]
bendlas has quit [Ping timeout: 240 seconds]
<toolchains_x86>
And if I hit an error in the compilation process, or in the Makefile, the process starts from the beginning every time. Which is a pain to work with.
<exarkun>
I don't understand what you're compiling or how pypy helps with rpython. Are you ... writing your own custom tool in RPython?
<exarkun>
er, helps with openwrt
<LarstiQ>
antocuni: hmm, do you have a link to look for that?
<LarstiQ>
exarkun: compiling pypy itself, afaics
<fijal>
toolchains_x86: then do a shortcut starting from a binary
<LarstiQ>
fijal, toolchains_x86: I'd suspect the problem is at the stage of the platform checks/how the host mismatches the target there?
<kenaan>
Dodan py3.5-sendmsg-recvmsg a8d5fdebbce7 /: Fix for the test_EINTR errors on sendmsg and recvmsg
<antocuni>
LarstiQ: not really, apart the ones I can find by googling "pypy nokia maemo"
<antocuni>
it seems it was pypy 1.1, so really old
yuyichao_ has quit [Ping timeout: 258 seconds]
<LarstiQ>
antocuni: for maemo's current successor Mer, it runs under qemu-arm with accelerators anyway
<kenaan>
mattip cpyext-refactor-tp_dealloc e3219c61e7a8 /: fix translation by adding immortal to malloc in rpython
<kenaan>
mattip default 54cc281fb04e /pypy/doc/: improve cross-referencing in documentation
<toolchains_x86>
exarkun: Yes, and I would prefer to increase the speed of running my Python.
<toolchains_x86>
fijal: How would I start from a binary?
<kenaan>
mattip cpyext-refactor-tp_dealloc 98384706fd2b /: close dead-end branch
<fijal>
toolchains_x86: compile it and have a binary
<fijal>
then write the packaging stuff
yuyichao_ has joined #pypy
cstratak has quit [Quit: Leaving]
cstratak has joined #pypy
<mattip>
exarkun: added links. More suggestions are welcome
<toolchains_x86>
fijal: I don't think I understand that too well, could you expound on it?
<fijal>
toolchains_x86: how does your build work?
adamholmberg has joined #pypy
yuvipanda has joined #pypy
adamholmberg has quit [Remote host closed the connection]
adamholmberg has joined #pypy
cstratak has quit [Quit: Leaving]
cstratak has joined #pypy
<mattip>
plan_rich: latest vmprof commits broke win32 translation and untranslated tests are failing
Ryanar has quit [Quit: Ryanar]
Ryanar has joined #pypy
bendlas has joined #pypy
toolchains_x86_ has joined #pypy
<toolchains_x86_>
fijal: What do you mean?
<mattip>
toolchains_x86: making a functioning PyPy distribution has at least these stages -
<mattip>
translating the rpython code to c, which is single process/single thread takes a long time
<mattip>
compiling the c code to an exe, via the Makefile
<mattip>
packaging the resulting executable + shared object + cffi extension modules + python code into a python distribution
<mattip>
ahh, and building the cffi extension modules
<mattip>
so you successfully build the makefile, that means you can save that directory and redo the compile step over and over
<mattip>
and if the makefile sucessfully runs, you can build the cffi extensions over and over until they work
<mattip>
and if that all works, you can package over and over until that works
<mattip>
if you find yourself backing up more than one step, you should think about why
* mattip
off, sorry
mattip has left #pypy ["bye"]
david-fb has joined #pypy
<david-fb>
Hello. I would like to configure the boost C++ libraries to generate valid pypy 3.5 extension modules. Is this possible? Any guidance on what is required? Thanks
<ronan>
david-fb: In principle, it should be no different than doing it for CPython. Just try it and see where it breaks.
<david-fb>
:-)
raynold has joined #pypy
<exarkun>
Consistent failure of test.test_asyncio.test_base_events.BaseEventLoopWithSelectorTests.test_create_datagram_endpoint_sockopts on buildbot but consistent success locally, hm.
<cfbolz>
yay
<cfbolz>
exarkun: you can probably get a login to the machine that runs the build, if you want to dig
<fijal>
yay asyncio
<exarkun>
I guess that might help, yea
<cfbolz>
fijal: can you make exarkun a bencher4 account?
marr has quit [Ping timeout: 255 seconds]
<fijal>
cfbolz: don't we need to ask someone etc?
<cfbolz>
fijal: why?
<fijal>
I don't know, don't remember procedures
<fijal>
sure
<cfbolz>
It's fine
_whitelogger has joined #pypy
toolchains_x86 has quit [Ping timeout: 260 seconds]
jcea has quit [Ping timeout: 240 seconds]
vkirilichev has quit [Remote host closed the connection]
jcea has joined #pypy
<pjenvey>
SO_REUSEPORT related
<pjenvey>
exarkun: I think that one is just a silly flag doesn't exist on that platform issue
<pjenvey>
exarkun: IIRC the 3.6 tests accomodate that flag not existing, but 3.5 never fixed it, might want to double check that
Fleche has quit [Quit: Fleche]
mattip has joined #pypy
oberstet has quit [Ping timeout: 260 seconds]
forgottenone has quit [Quit: Konversation terminated!]
cstratak has quit [Ping timeout: 255 seconds]
_zware has joined #pypy
<exarkun>
What's the eventual goal for the py35 branch? Be merged to default, making PyPy a 3.x runtime? Live alongside default indefinitely so there are PyPy 2.x and 3.x runtimes?
<fijal>
the latter
<fijal>
exarkun: well, about 15% of downloads from pypi are for python 3
<fijal>
it's not exactly a success story
antocuni has quit [Ping timeout: 240 seconds]
<exarkun>
So if there's, eg, objspace changes to make for 3.x compat ... just make them, even if they're not 2.x compatible
<fijal>
yes, there is no real way around it
<fijal>
we *try* to minimize the diff
<fijal>
so one thing we don't do is to modify rpython dir for py3k