cfbolz 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 | if a pep adds a mere 25-30 [C-API] functions or so, it's a drop in the ocean (cough) - Armin
RemoteFox has quit [Remote host closed the connection]
RemoteFox has joined #pypy
jcea has quit [Remote host closed the connection]
jcea has joined #pypy
jcea has quit [Remote host closed the connection]
_whitelogger has joined #pypy
jcea has joined #pypy
_whitelogger has joined #pypy
jcea has joined #pypy
dddddd has quit [Remote host closed the connection]
jcea has quit [Remote host closed the connection]
jcea has joined #pypy
jcea has quit [Quit: jcea]
lritter has joined #pypy
jvesely has quit [Quit: jvesely]
dmalcolm_ has joined #pypy
dmalcolm has quit [Ping timeout: 245 seconds]
marky1991 has quit [Ping timeout: 276 seconds]
tazle has quit [Read error: Connection reset by peer]
altendky has quit [Quit: Connection closed for inactivity]
<antocuni> uhm, now in the hpy branch we have module/hpy_universal/_vendored, which contains files copied from pyhandle/hpy, including test/ (which are then imported by hpy_universal/test/*.py to run them at applevel). However, pytest will ALSO try to run _vendored/test and fail
<antocuni> any good idea to prevent it? Preferably not involving renaming the files
<antocuni> maybe some conftest hacking? Or is there a way to tell pytest "please don't collect this dir"?
<arigato> I think it's done with conftest hacking, yes
<antocuni> then the question is... where should the conftest be? in hpy_universal? inside _vendored? (a bit obscure since it's not a vendored file)
<antocuni> or some level up?
<antocuni> arigato: also, unrelated: see my XXX in hpy's commit d0718ef :(
<antocuni> we need a way to solve it
<antocuni> (currently trying to implement HPyArg_Parse in pypy)
<arigato> yes
<antocuni> btw, speaking about this: I am copying&pasting ctx_Arg_Parse into src/getargs.c. But then in order to reference it from RPython and stick it into the ctx, I need it to be declared as non-static, I think :(
oberstet has joined #pypy
lritter has quit [Quit: Leaving]
<arigato> in pypy, you can use RPY_EXTERN if you include "commonheader.h" I think
<arigato> it makes a symbol non-static, but not exported
<arigato> only visible from the other .c sources that will be linked together to make the final .so
<kenaan> arigo default dcc4efe7355e /pypy/module/cpyext/: Rename 'tuple_new' to '_PyPy_tuple_new' to follow the naming convention of exported symbols in libpypy-c.so.
YannickJadoul has joined #pypy
yusuke_izawa has joined #pypy
yusuke_izawa has quit [Client Quit]
yusuke_izawa has joined #pypy
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
yusuke_izawa has quit [Ping timeout: 240 seconds]
yusuke_izawa has joined #pypy
yusuke_izawa has quit [Quit: Leaving.]
yusuke_izawa has joined #pypy
<kenaan> antocuni hpy f9c2cfcfaafa /pypy/module/hpy_universal/: This checkin might win the prize for the highest amount of XXXs/lines of code: it needs a deep review, please :) Fi...
yusuke_izawa has quit [Ping timeout: 240 seconds]
Rhy0lite has joined #pypy
xcm has quit [Read error: Connection reset by peer]
xcm has joined #pypy
yusuke_izawa has joined #pypy
yusuke_izawa has quit [Client Quit]
mattip_ has quit [Ping timeout: 245 seconds]
mattip_ has joined #pypy
mattip_ has quit [Client Quit]
<Dejan> Did anyone here compile PyPy 3.6 with musl C ?
<Dejan> Actually, I am thinking about building statically linked pypy 3.6 and make my base Docker container out of it (from scratch)
<hotpot33> Dejan: hmm how does the docker pypy do it
<hotpot33> isn't it also based on alpine?
<Dejan> no, i basically want to statically link pypy and make a docker image "FROM scratch"
<Dejan> "FROM scratch" is basically a NOOP
<Dejan> alpine is small, but quite often I really do not need busybox etc there
<Dejan> it explains the basic idea
<Dejan> the reason why i ask about musl C is because glibc static library is enormous , while muslC is ~2Mb
<Dejan> antocuni, how difficult would it be to have fastparquet in your manylinux2010 repo?
<Dejan> :)
<antocuni> Dejan: the first question is whether it works. I see it depends on numba, I don't know if anyone has ever tried it with pypy
<Dejan> ah yea... forgot it needs numba...
<antocuni> it might work
<antocuni> I simply don't know :)
<Dejan> actually, i could switch to pyarrow
<Dejan> blah, pyarrow failes to build because CMake's FindNumPy can't find numpy
<Dejan> i will try this...
jvesely has joined #pypy
YannickJadoul has quit [Quit: Leaving]
dddddd has joined #pypy
<ronan> Does anyone understand what's going on with the failures in extra_tests/cffi_tests on bencher4?
mattip_ has joined #pypy
<mattip_> ronan: i cleaned out the /tmp dierctory
<mattip_> not that it was full, but maybe there were too many nodes
mattip_ has quit [Remote host closed the connection]
<kenaan> rlamy py3.6 fc3e8b80874c /pypy/module/sys/test/test_sysmodule.py: Remove obsolete test
danyspin97 has joined #pypy
mwhudson has quit [Ping timeout: 250 seconds]
<danyspin97> will rpython code be ported to python3?
<simpson> danyspin97: AIUI the plan is no, instead PyPy is self-hosting and can run RPython.
<ronan> danyspin97: maybe, but it's a lot of work
<danyspin97> because if there is no pypy3 to use for bootstraping, then python2 is a dependency
tazle has joined #pypy
<ronan> danyspin97: yes, that will become an issue eventually
<danyspin97> ronan: python2 will be discontinued in a month, it could be considered an issue :/
<simpson> danyspin97: CPython only. PSF cannot unilaterally deprecate PyPy's Python 2 support.
<ronan> danyspin97: end of support doesn't mean that cpython2 will suddenly stop working
<simpson> The eventually-issues are things like PyPI support, package incompatibilities, that sort of secondary effect.
<danyspin97> simpson: without a bootstrap pypy (i.e. when using musl), CPython2 (deprecated) is needed to get a pypy binary
<danyspin97> using something deprecated into the toolchain could be considerated an issue imho
<cfbolz> danyspin97: we need at least a few person years to port, that money needs to come from somewhere
altendky has joined #pypy
<simpson> danyspin97: Hmmm. Is that really the case, that PyPy can't cross-compile from one libc to another? Unfortunate if so.
<simpson> I hear your larger issue, but like cfbolz and ronan say, it's a lot of expensive work.
dansan has quit [Ping timeout: 250 seconds]
dansan has joined #pypy
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
hsaliak has joined #pypy
oberstet has quit [Remote host closed the connection]
Rhy0lite has quit [Quit: Leaving]
mattip_ has joined #pypy
<mattip_> how is using an old cpython different from using an old gcc? gcc4.8 was last updated in 2015
mattip_ has quit [Remote host closed the connection]
mwhudson has joined #pypy
mwhudson has quit [Changing host]
mwhudson has joined #pypy
Ai9zO5AP has joined #pypy
i9zO5AP has quit [Ping timeout: 240 seconds]
<danyspin97> simpson: yea, then we could not do a lot about that
<danyspin97> thanks anyway ^^
<simpson> danyspin97: FWIW I've been thinking about the issue a *lot*, and paying attention on IRC the past couple days.
<simpson> I'd love to be able to say that nixpkgs has PyPy working on musl, but we don't. Patches would be welcome. It will take a lot of patience to try building PyPy many times, though.
<danyspin97> simpson: I'll try tonight, bootstraping with python2
<danyspin97> the distro I am using had already pypy and pypy3 packages in the past, I'm bumping them right now
<simpson> Ah. Good luck.
<danyspin97> `ctypes.util.find_library('c')` == None
<danyspin97> surely not a great start
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
inhahe_ has quit []
inhahe has joined #pypy
jcea has joined #pypy