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
abrown has joined #pypy
<kenaan>
Yannick_Jadoul py3.7-pep564 f0362a20d1a1 /: Adding ns-resolution function to time module
<kenaan>
Yannick_Jadoul py3.7-pep564 fb46c09b060b /pypy/module/time/interp_time.py: Docstrings for time functions
<kenaan>
Yannick_Jadoul py3.7-pep564 7798b261cc60 /pypy/module/time/test/test_time.py: Added tests for nanosecond-resolution time functions
<kenaan>
Yannick_Jadoul py3.7-pep564 9741d65e2e03 /pypy/module/time/interp_time.py: Fix return type of edge-case of gettimeofday when only C's time() is available
<mattip_>
arigato: thanks. Maybe we should add a comment somehow to pyopcode that class creation moved to __pypy__.compiling.py
<arigato>
that's clear enough I think
mattip_ has quit [Ping timeout: 268 seconds]
jacob22_ has joined #pypy
Guest90013 has quit [Ping timeout: 268 seconds]
antocuni has quit [Ping timeout: 240 seconds]
Guest90013 has joined #pypy
Guest90013 has quit [Quit: Leaving]
ronan has quit [Ping timeout: 268 seconds]
<arigato>
*of course* the manylinux build of cffi 1.13.1 is broken because it contains its own version of libffi.so, and that version is buggy
<arigato>
$ yum install -y libffi libffi-devel
<arigato>
that seems to install a buggy version
bernte has joined #pypy
<tumbleweed>
yay :/
<arigato>
I imagine there's no way to just use "yum" and get a more recent libffi?
<arigato>
for now I'm fighting all the way into downloading and compiling libffi from sources, but it's a nightmare too
<arigato>
it seems I'll need to yum install the whole large TeX packages
* tumbleweed
doesn't know enough about the RH world :(
<arigato>
I love how setting up azure pipelines feels like a high-flying waste of resources
<arigato>
you tweak some shell script, commit it, go to the web page, see that it contains a typo, restart---so far OKish, but every time you do that, there's a complete docker VM being downloaded, and there's also OSX and Windows builds running to completion
<arigato>
Windows builds in particular require downloading GBs of VMs
<arigato>
so, every typo costs several GBs
<arigato>
inot a waste at all
YannickJadoul has joined #pypy
<YannickJadoul>
arigato: This is manylinux2010/CentOS 6, right? So in principle, it's still supported?
<YannickJadoul>
There were some concerns about yum installing things on manylinux1, since these packages would be old and potentially without the latest security patches
<Alex_Gaynor>
FWIW when we were building cffi wheels we built our own libffi
<YannickJadoul>
On debugging CI configs: my condolences. It's a real pain. I'm still wondering if using CIs actually saved me time in the long run or not...
<arigato>
Alex_Gaynor: bah. Yes, I'm also trying to install the "recent" version (2014) of libffi manually. Seems to work now
<arigato>
this complete mess of distributing wheels certainly doesn't make me happy
<arigato>
I will certainly look twice before doing more changes to cffi