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
<cfbolz>
YannickJadoul: hey! I was wondering, do you plan to work on the positional only argument parsing soon? Otherwise I might continue to work on that branch, if that's OK with you?
tsaka__ has quit [Ping timeout: 272 seconds]
xcm has quit [Ping timeout: 255 seconds]
xcm has joined #pypy
<YannickJadoul>
cfbolz: Yeah, I'd like to get back to that, but please go ahead if you want!
<YannickJadoul>
The main problem I had is that the error messages are not that great to solve the `int(x=42)` failure
<cfbolz>
ok, might have some time to take a look at it, hopefully next week
<cfbolz>
YannickJadoul: yes, there are a few cases where the error messages around calls are worse in pypy than cpython right now
<YannickJadoul>
Yeah, so it seems CPython has 2 different ways of reporting them, as well
<cfbolz>
yep
<YannickJadoul>
`TypeError: 'x' is an invalid keyword argument for int()` vs. `TypeError: f() got some positional-only arguments passed as keyword arguments: 'a'`
<YannickJadoul>
So I think I got a rough sketch of 3.8 positional-only arguments working
<YannickJadoul>
But I wasn't sure it was the right way forward of solving the `int(x=42)` in Python 3.7? :-/
<YannickJadoul>
(it's a bit weird to give errors in 3.7 about a feature that's only present in 3.8, no?)
<cfbolz>
yes, a bit
<cfbolz>
but also annoying to write code that will then have to be changed again in
<cfbolz>
3.8
<YannickJadoul>
Yeah, but well, I checked that: even in 3.8, `int(x=42)` gives the same error as in 3.7, without mentioning positional-only
<YannickJadoul>
That error is somehow coming from the argument clinic, iirc
<cfbolz>
yes, CPython parses arguments differently whether they are to builtin functions vs python-defined ones
<YannickJadoul>
Quite annoying, yes. So the current plan was to see if there was another way, then make a PR comparing the two
<YannickJadoul>
Then I got caught up in other stuff; sorry :-/
<tos9>
YannickJadoul: (on the virtualenv thing -- I would not count myself as "knows", certainly "interested" though, and I am almost certainly in the top 1% of usage of virtualenv+pypy for personal development so I run into lots of fun issues like that one -- will probably follow the ticket -- although you know gabor is rewriting virtualenv, so he'll probably be pretty engaged)
<YannickJadoul>
tos9: Thanks :-) I'm just a bit horrified by the combo of macOS + virtualenv 20 + PyPY 3.6 + pytest; still trying to figure out the minimal way of reproducing it, so I'm sure that any help or ideas are welcome :-)
<tos9>
YannickJadoul: do you have a non-minimal way I should try? (if it's in the ticket I can look in a bit)
tos9 has quit [Read error: Connection reset by peer]
tos9 has joined #pypy
<YannickJadoul>
tos9: What do you mean with non-minimal?