cfbolz changed the topic of #pypy to: #pypy PyPy, the flexible snake https://pypy.org | IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end and https://libera.irclog.whitequark.org/pypy | hacking on TLS is fun, way more fun than arguing over petty shit, turns out
sbrown2 has quit [Quit: Leaving…]
glyph has quit [Quit: End of line.]
glyph has joined #pypy
<fijal> do we know when did python start checking certificates?
<mattip> fijal: are you having problems with hg? It seems if you update certifi it fixes the problem
<fijal> no, I wonder if a slowdown from pypy6.0 is due to not checking certificates back then
<fijal> hg does not work indeed, but that's abnother story
<mattip> not only hg. The benchmarker is running in a xenial chroot (ubuntu 16.04) with python2.7.11, and cannot upload results to speed.pypy.org
<mattip> there must be a way to get rid of the expired certificate from the bundle in /etc/ssl/certs/ca-certificates.crt
<LarstiQ> mattip: normally that would be updating ca-certificates package, but I suppose there are no updates for that. Afaik it's `update-ca-certificates` to make use of new certs you drop in
<mattip> right, there are no updates
<mattip> I could update openssl to 1.1.1, but I think the version of python is too old to use it
<cfbolz> mattip: weird. I'll try to take a look soon
<fijal> arigo_: do you feel like a performance quiz?
<fijal> this is the benchmark, guess relative timings
<fijal> ok....
<fijal> pom pom pom pom pom
<mattip> ?
<fijal> mattip: the ssl sockets are a complete mess
<mattip> makes sense. I did the minimum needed to pass tests
<fijal> well, it's a bit bad for us I think - we have no SSL benchmarks, so if it's bad it's going to be bad
<fijal> but also, *everyone* uses SSL these days, so it seems like it's an important benchmark
<mattip> ok. Is there something we can use for a benchmark?
<mattip> it seems we need to recalibrate the benchmarks anyway, the runner is failing and the machine needs to update the kernel
<fijal> ping-pong in ssl would be a start?
<fijal> I'm doing some boto and some redis (they're both a bit bad for a benchmark)
<fijal> I'm not volunteering to set up certificates correctly
<cfbolz> mattip: ok, so the test_code_module test failure is the test does just not expect any suggestion at all
<cfbolz> code uses the traceback module
<cfbolz> which in cpython does *not* give suggestions
<cfbolz> (the suggestions are in C only)
<cfbolz> so I am going to fix the test
<mattip> cool, thanks for looking
<cfbolz> hmmmm, something else is weird though
<cfbolz> my cpython3.9 doesn't seem to give me any suggestions
<mattip> that's because it is new to 3.10
<mattip> but we backported it to 3.8
<cfbolz> Ahhhh
<cfbolz> Thank you
<cfbolz> I was very confused for a bit
<mattip> pypy3.8 passed the numpy test suite on windows and linux x86_64
<cfbolz> nice!
<cfbolz> extremely cool
<cfbolz> mattip: that's the first time, right?
<cfbolz> (there's definitely something wrong with suggestions, btw)
Atque has quit [Quit: ...]
<LarstiQ> mattip: woo!
the_drow has quit [Ping timeout: 250 seconds]
<mattip> it is not the first time I tried with NumPy locally, the first I tried to make a PR
<cfbolz> cool
the_drow has joined #pypy
<mattip> it led to the whole os.add_dll_directory() thing for windows that I forgot about for 3.8
jryans has quit [Ping timeout: 264 seconds]
jryans has joined #pypy
greedom has joined #pypy
<mattip> do we have a way of doing a bugfix release after 7.3.6, or would that have to be 7.3.7?
<mattip> we need a new release of virtualenv for pypy3.8,
<mattip> but there is some subtle problem with virtualenv + windows blocking the virtualenv release
<mattip> I would like to release 7.3.6 and fix the issue later, but not if I have to update to 7.3.7
<tumbleweed> does the machinery support micro-versions? 7.3.6.1
greedom has quit [Ping timeout: 246 seconds]
greedom has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
greedom has quit [Remote host closed the connection]
greedom has joined #pypy
greedom has quit [Remote host closed the connection]
<cfbolz> mattip: ok, pushed a fix for test_code_module
<cfbolz> there are some differences for our suggestions to CPython's, but can be fixed later
directive0 has joined #pypy
<directive0> Hi. I have a really stupid question that I can't seem to find an answer for: I want to use pypy3.7 on a raspberry pi. I have seen some suggestions its supported, but don't see any binaries for this purpose. Is this actually something supported at this time and I just have to build it from source, or am I completely misguided
commandoline has joined #pypy
commandoline has quit [Client Quit]
commandoline has joined #pypy
directive0 has quit [Quit: Client closed]
stkrdknmibalz has joined #pypy
Atque has joined #pypy
ceridwen has joined #pypy