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
<mattip>
sent a mail to pypy-dev about starting a release cycle, help is welcome
<mattip>
especially someone with a mac, if they could figure out why default can run lib_pypy/_ssl_build.py and py3.6 cannot, and can tell me what version of libressl we should be using
<mattip>
(we download one and ship it with pypy on macos)
<the_drow>
This is PyPy 7.1.0-beta0 with GCC 8.2.0
<the_drow>
Python 3.6
<the_drow>
I'm going to try pypy 7.1.1 and see if that makes a difference
<tos9>
mattip_: what am I trying to figure out here, sorry -- just try to execute that file on the 2 branches?
<mattip>
tos9: yes, then see what <pypy> -c "import ssl; print(ssl.OPENSSL_VERSION)" gives you
<mattip>
at least on pypy2 where it should work
<tos9>
the pypy3 nightly gives... AttributeError: cffi library '_pypy_openssl' has no function, constant or global variable named 'Cryptography_HAS_ECDH'
<tos9>
That's what we're tryign to fix?
<mattip>
right. What I don't understand is what is the difference with pypy2 where that passes
<tos9>
Uh that was also with homebrew openssl by the way.
<tos9>
Without homebrew openssl I got failures looking for the openssl headers.
<tos9>
mattip: OK yeah but that succeeds on pypy2
<tos9>
And gives me the openssl I told it to use ('OpenSSL 1.0.2s 28 May 2019', which is the current version in homebrew)
<mattip>
we could put homebrew into the build_cff_imports.py, but I think that would be overkill for what we need
<tos9>
mattip: I think what I meant was "why doesn't this script just assume someone should have openssl installed externally via homebrew, instead of fetching it itself", not that it should install homebrew itself
<tos9>
but... whatever, first -- looks like it doesn't just run directly for me quite yet
<tos9>
and first I need to figure out hg sigh... how do I delete a branch
<tos9>
or get it to not push this branch remotely
<tos9>
or tell push to only push one branch
<tos9>
ok hg push takes -b..
<mattip>
if u use thg, you can right click on the base commit and mark it "private", or you can use the gui "detect outgoing changes" and then it will shou you what it wants to push
<mattip>
mark only the ones you want and push them
<mattip>
tos9: cool, thanks
<tos9>
mattip: I didn't do anythign yet :)
kipras has joined #pypy
kipras has quit [Read error: Connection reset by peer]
<ronan>
cfbolz: AFAICT CPython 3.8 deals with missing POP_BLOCK correctly, but not with missing END_FINALLY
<cfbolz>
ronan: right
<ronan>
but they err on the side of allowing wrong jumps :-(
jcea has joined #pypy
<ronan>
I guess the only way to make it correct is to reconstruct exactly the cases where the compiler omits some opcodes
jcea has quit [Remote host closed the connection]
tsaka__ has quit [Ping timeout: 245 seconds]
oberstet has joined #pypy
ronan has quit [Ping timeout: 246 seconds]
BPL has joined #pypy
lesshaste has joined #pypy
jcea has joined #pypy
<tos9>
mattip_: OK that seems to "work" at least inasmuch as giving the new version now correctly: https://bpaste.net/show/D8LI
<tos9>
(the 2 commits I pushed, which like you said just change those lines)
<tos9>
mattip_: If I kick a buildbot on that branch, based on what you said before I assume it should try to run the tests with that openssl version right?