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
jacob22 has quit [Ping timeout: 252 seconds]
CrazyPython has joined #pypy
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
CrazyPython has quit [Read error: Connection reset by peer]
<kenaan>
arigo cffi/cffi 32b6ed4cb029 /: Add a warning when we use in cdef() a global variable without also specifying a storage class (extern or static)
<kenaan>
arigo fix-descrmismatch-crash 8c25c3e5ab7d /pypy/: frozenset.__init__() does nothing on CPython, and can be called with random arguments.
jacob22 has joined #pypy
<kenaan>
arigo fix-descrmismatch-crash 9acddd120236 /pypy/interpreter/: Make a new helper method for the common code, and move there the logic that should safely fetch the...
forgottenone has joined #pypy
<kenaan>
arigo fix-descrmismatch-crash 78855fca6aa9 /pypy/: Improve the error message by using the class name 'set-or-frozenset'
<arigato>
cfbolz, mattip: I've found a better way to fix the name 'internal subclass of W_Xyz' than to go ahead and change all interp*2app(), which is probably wrong in general
<kenaan>
arigo fix-descrmismatch-crash c13fae180ec6 /pypy/objspace/std/: fix the error message also for W_AbstractBytesObject
<kenaan>
arigo fix-descrmismatch-crash cc73504e9f94 /pypy/module/_weakref/: same fix for weakref-or-proxy
<kenaan>
arigo fix-descrmismatch-crash 56cde2db6b50 /pypy/module/cStringIO/: Same fix for cStringIO
<kenaan>
arigo fix-descrmismatch-crash c2572a2d1270 /pypy/objspace/std/test/: Some more cases of W_AbstractXxx but where the test already passes
<kenaan>
arigo py3.6 3c57a384aae3 /pypy/objspace/std/bytesobject.py: the merge of 'fix-descrmismatch-crash' into 'py3.6' will need this line, not 'str'
<cfbolz>
arigato: the whole mechanism is a bit weird, it was invented for transparent proxies, I think
<arigato>
yes
<arigato>
I think it should always call the call_descr_mismatch method with the first positional argument
<arigato>
in one case it was using the first argument after 'self'
<cfbolz>
arigato: yes, both parts look much nicer
jerith_ is now known as jerith
<cfbolz>
I mean both exact_class_applevel_name and the change in gateway
CrazyPython has quit [Read error: Connection reset by peer]
<altendky>
mattip: I try to help out projects getting ci and wheel builds setup and I like to include PyPy support. Though I can't say I've read up on supporting that properly. Is there something I should read for that?
<mattip>
altendky: do they use cython or the C-API? creating c-extensions is tricky since
<mattip>
1) we would prefer pure python
<mattip>
2) or at most cffi
<mattip>
3) and if they have to use c-extensions, they should be using a project like manylinux to make sure the wheels support most glibc versions
<altendky>
mattip: it depends. I've used cibuildwheel to manage the manylinux aspect for those that need to compile. I would encourage cffi over the cpython c-api but... Not going to rewrite entire libs usually.
<mattip>
but it uses portable-pypy which I don't think has 7.2 yet
<altendky>
mattip: I'm happy to read if there's a link and honestly haven't really looked... But how many wheels and what varies between them to get 'full coverage' for PyPy? Like for cpython I would do win/Linux/Mac * 2.7/3.5/3.6/3.7(/3.8). Though maybe that's silly when it comes to cffi... Still plenty for me to learn.
<mattip>
win/x86 linux/mac 2.7/3.6 is about all you can do currently on PyPI. manylinux2014 will support arm and aarch64, but it is not done yet
<mattip>
travis has pypy 7.1, maybe we could encourage them to install 7.2
<altendky>
mattip: i mostly go for azure pipelines at this point.