<Ninpo>
Any way I can troubleshoot a spot where pypy3.5 hangs and all strace -p gives is futex(0x7f1e31def000, FUTEX_WAIT_BITSET|FUTEX_CLOCK_REALTIME, 0, NULL, 0xffffffff ? It doesn't happen every run of code and cpython hasn't exhibited the behaviour thus far.
<fijal>
Ninpo: do you know what kind of module can cause it?
<fijal>
e.g. are you using multiprocessing? cffi?
<fijal>
some C extension?
<fijal>
Ninpo: generally reducing the example to something we can run is a great step
<fijal>
if we can run it, even if it's not very minimal, we can usually debug it
<fijal>
mattip: ^^^
<mattip>
fijal: there is a numpy dev who wishes to come Tues night, can we put him up?
<fijal>
mattip: I already responded to your email :) do you want to see the place first?
<Ninpo>
let me quickly create a pypy branch for this app. It fetches rows from a mysql server and processes text in them.
<fijal>
Ninpo: please create an example and reproducible instructions
<fijal>
e.g. creating a mysql database with some payload is fine (but oracle one is not for example)
<Ninpo>
When it doesn't hang it's WAY faster than the cpython run. I've been working on RAM usage reduction, but sometimes yeah, pypy just hangs.
<Ninpo>
My first "proper" attempt at an app so please if there's anything glaringly stupid feel free to hit me over the head with it :D
<fijal>
Ninpo: ok, we need a way to reproduce it right?
<fijal>
right now I see a repo with a bunch of py files, how do I run it?
<fijal>
the steps are as follows:
<Ninpo>
To get the app running, pip install the git path. Not sure how to get around needing a mysql server to try and get it to hang :/
<fijal>
installing mysql server is fine
<fijal>
having it not mentioned anywhere is not
<Ninpo>
That's why I initially asked if there was anything I could do to get more useful info out
<fijal>
please do the following:
<fijal>
1. please write down instructions how we can run it, assume we know nothing about your app
<Ninpo>
The README does say it's for mysql....but yeah I'll update the README
<fijal>
2. please create a ticket on pypy issue tracker with necessary links
<fijal>
"Identify and fix mojibake problems in MySQL-like RDBMS" <- to me that says nothing with "to run please install mysql server and put your credentials HERE"
<Ninpo>
Fair enough. I'll get something together once ruled out anything stupid my side. Just wasn't sure if there was anything unique to pypy for getting debug info out :) Thanks fijal
<Ninpo>
I love this project, most of the things I work on are long running, purely python and munch on a lot of stuff, the performance is incredible.
<fijal>
there are many pypy-specific things, but it's a bit unlikely we can do anything without having a debugger running
<fijal>
"bit unlikely" - should really say "impossible"
<Ninpo>
While I sort some test data and better instructions, I did get this after a number of CTRL+Cs, no idea if it's useful to you
<antocuni>
arigato, mattip: I merged default and py3.5 into the release branches, to include the latest cffi changes
<antocuni>
what about py3.6? I don't see any recent cffi change in the py3.6 branch, but I'm not sure what is the correct thing to do in order to do it
<antocuni>
I suppose it is: merge 3.5 into 3.6, and then 3.6 into release-pypy3.6-7.x, but before messing things up I prefer to wait for confirmation :)
marky1991 has joined #pypy
<antocuni>
I am about to leave soon; in case it is the correct thing to do, please do the merges and run ./force-builds.py -b release-pypy3.6-7.x
<arigato>
antocuni: yes, as far as I know merging 3.5 into 3.6 is fine
<arigato>
I can do the merges
<antocuni>
arigato: I'm still here; have you done them already or should I?
<arigato>
it's not a bug, it's a feature! it allows for line continuations: if you write "^@\n" instead of just "\n" at the end of a source file line, then the next line will be considered part of it
agronholm_ has joined #pypy
agronholm_ has quit [Read error: Connection reset by peer]
agronholm has quit [Read error: Connection reset by peer]