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
<nimaje>
ok, trying to translate pypy with jython ends pretty fast sys.version_info is a plain tuple on jython https://0x0.st/-iXN.txt (well ensurepip also doesn't work and jython doesn't bring cffi by default, so it would have failed later anyway)
<LarstiQ>
nimaje: hmmm, one would expect that to be a more widely impactful divergence of jython?
marvin_ has quit [Remote host closed the connection]
marvin_ has joined #pypy
<mattip>
it seems pickle5 is very easy to support: we just need to add a few lines to the pure-python pickle.py implementation, and update some constants
<mattip>
hmm, deleting the _pickle c-extension module from cpython makes the sphinx benchmark go from 41 secs to 61 (pypy3.7 and 2.7 are both ~85 secs)
<mattip>
I wonder if an rpython _pickle module would speed things up
commandoline_ has joined #pypy
marvin_ has quit [Ping timeout: 256 seconds]
commandoline has quit [Ping timeout: 256 seconds]
marvin___ has joined #pypy
lobo_ has quit [Ping timeout: 256 seconds]
lobo_ has joined #pypy
Marco91 has joined #pypy
isidentical has quit [Read error: Connection reset by peer]
isidentical has joined #pypy
dansan has quit [Ping timeout: 246 seconds]
dansan has joined #pypy
<cfbolz>
mattip: doesn't the cpython pickle.py implement pickle5 correctly? ie don't we automatically pick up the changes?
isidentical has quit [Read error: Connection reset by peer]
isidentical1 has joined #pypy
isidentical1 is now known as isidentical
isidentical has quit [Client Quit]
<mattip>
for python3.8 there is no problem. There are projects that run faster when using pickle5 on 3.7
<mattip>
but in general I am still trying to think of a strategy to speed up sphinx, I thought maybe pickle could help
<cfbolz>
mattip: right
<cfbolz>
mattip: maybe we can put time calls around the pickle entry points and get the time we spend in there?
<cfbolz>
(but indeed, we already have signals that pickling is a problem in the actual pickle benchmarks)