arigato 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 | mac OS and Fedora are not Windows
ronan has quit [Ping timeout: 252 seconds]
ronan has joined #pypy
speeder39_ has joined #pypy
jcea has quit [Remote host closed the connection]
_whitelogger has joined #pypy
tsaka_ has joined #pypy
tsaka__ has quit [Ping timeout: 244 seconds]
lritter has quit [Ping timeout: 268 seconds]
lritter has joined #pypy
_whitelogger_ has joined #pypy
speeder39_ has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #pypy
<mattip> Alex_Gaynor: shouldn't be too hard, you need the test_ssl.py file and its support functions, and the _stdssl directory
altendky has quit [Quit: Connection closed for inactivity]
inhahe has quit []
inhahe has joined #pypy
speeder39_ has joined #pypy
<wleslie> I feel rather out of the loop that I had not heard about mesapy until today
<wleslie> I mean - I expected that someone would attempt to do memory hardening with pypy sooner or later, but I hadn't even heard about the attempt, and now it's a thing
<wleslie> similarly with formal verification of the JIT
antocuni has joined #pypy
antocuni has quit [Remote host closed the connection]
antocuni has joined #pypy
<cfbolz> wleslie: I am a bit unclear what they are doing
<cfbolz> we already have memory hardening built into pypy
<cfbolz> it's what you get when you compile with "make lldebug"
<cfbolz> (I suppose you want just a subset of that target, but one of the things that the target enables is bounds checking on everything)
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
ronan has quit [Quit: Ex-Chat]
ronan has joined #pypy
antocuni has quit [Ping timeout: 272 seconds]
speeder39_ has quit [Quit: Connection closed for inactivity]
altendky has joined #pypy
<mattip> there is a long discussion on the cpython discourse about reimplementing the stdlib in pure python
<mattip> I find that both that discussion and projects like mesapy, and the pypy sandbox ignore the main problem:
<mattip> ongoing support for changes and new versions
<mattip> maybe we have solved that problem for the sandbox going forward, but just look how much work it is to upgrade pypy
<mattip> or how much work goes into a cpython release
<mattip> mesapy would need to redo the static analysis for each new version of pypy, and who would update the rust versions of stdlib modules?
ronan has quit [Ping timeout: 252 seconds]
<cfbolz> mattip: that's generally an open research question in software correctness proofs, however
<cfbolz> (ie how to deal with changes)
ronan has joined #pypy
<mattip> I was more commenting on how much work it takes to keep a python implementation going, after the massive once-off effort to get it started
<cfbolz> mattip: yes, that's definitely true
lritter has quit [Ping timeout: 258 seconds]
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
dddddd has joined #pypy
fling is now known as bedflinger
bedflinger is now known as fling
xcm has quit [Remote host closed the connection]
antocuni has joined #pypy
xcm has joined #pypy
<mattip> aarch64 translation passes on default, but fails on py3.6 with a "AssertionError: live GC var around v4792 = gc_thread_run()!"
<mattip> the slave used pypy2-v7.2-6885d94b89cf to translate, I will try to update to the latest nightly
<mattip> in the probably misplaced hope that something significant changed
Rhy0lite has joined #pypy
<simpson> Hi. I'm examining a downstream distro issue: https://github.com/NixOS/nixpkgs/issues/39356 The distro's Python maintainer desires PYTHONHOME, but PyPy doesn't have it; is there an obvious better/portable way to do it?
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
exarkun has joined #pypy
<Alex_Gaynor> mattip: Cool! in terms of adding tests https://github.com/pyca/cryptography/tree/master/.travis/downstream.d is the place to look
<mattip> simpson: it seems to be equivalent to setting PYTHONPATH (or prepending the desired path to an existing PYTHONPATH)
<simpson> mattip: Makes enough sense, thanks.
<mattip> Alex_Gaynor: I suggested another idea in the PR: create a _ssl repo under pyca
<Alex_Gaynor> mattip: if we have a copy of the tests then we also need to update them whenever pypy does -- I think our existing strategy for running downstream tests is more flexible
<mattip> Alex_Gaynor: not sure I understand. The install stage would clone the pypy repo and update the internal copy of crpytography, the test stage would run the test_ssl.py tests?
<Alex_Gaynor> mattip: yes, exactly like that
<mattip> interesting idea. The opposite idea (PyPy would vendor a pyca/_ssl module with tests, like it does with vmprof and cffi)
ekaologik has joined #pypy
<mattip> might encourage other developers to pitch in and improve it independent of PyPy
<mattip> I am not so in-tune of what is happenning in that area, is the stdlib ssh module considered to be high-quality?
<mattip> s/ssh/ssl/
<mattip> in both cases the test run would be added to the download.d repo
ekaologik has quit [Read error: Connection reset by peer]
<Alex_Gaynor> mattip: the stdlib ssl module is pretty good now, it used to be quite bad
antocuni has quit [Ping timeout: 246 seconds]
mattip has quit [Ping timeout: 244 seconds]
mattip has joined #pypy
BPL has joined #pypy
<kenaan> mattip cryptograhpt-2.7 401d69215728 /lib_pypy/_cffi_ssl/_stdssl/error.py: fix last failing test - add lower case error message stanza
<kenaan> mattip cryptograhpt-2.7 7ddff8a0053b /pypy/doc/whatsnew-head.rst: close branch to be merged
<kenaan> mattip default 5921676473ae /: merge cryptograhpt-2.7 which updates vendored cryptography to 2.7
<kenaan> mattip py3.6 168ba5471494 /: merge default into py3.6
<kenaan> mattip py3.6 c9b561590324 /: tweak test, add missing get_cipher
<kenaan> mattip py3.6 1deb84428ab6 /extra_tests/ctypes_tests/test_cast.py: add failing ctypes test that causes socket.sendall() failure
* arigato runs an interactive "rpython" to try to reproduce the crash of pypy-c-jit-linux-aarch64 Build #11 on py3.6
xcm has quit [Remote host closed the connection]
xcm has joined #pypy
BPL has quit [Quit: Leaving]
xcm has quit [Ping timeout: 272 seconds]
xcm has joined #pypy
Rhy0lite has quit [Quit: Leaving]
<arigato> mattip: of course re-running the same build succeeds
<arigato> I copied the PATH from the buildbot and ran the same command
<arigato> my bet is that if we just re-run the buildbot it will pass
<mattip> cool. I did update the pypy in that path from pypy2-v7.2-6885d94b89cf to pypy-c-jit-97168-f95a2bad0572-aarch64
<mattip> about 8 hours ago
<mattip> maybe pypy2-v7.2-6885d94b89cf was miscompiled
lritter has joined #pypy
<kenaan> mattip py3.6 e0689d0f47c6 /: test, fix subtle logic bug for memoryview.cast (when view.format is not 'B')
<mattip> i had to stare at cast_to_1D in cpython way too long to see what was different
speeder39_ has joined #pypy