cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | the secret reason for us trying to get PyPy users: to test the JIT well enough that we're somewhat confident about it
adamholmberg has quit [Ping timeout: 252 seconds]
ciphergoth has joined #pypy
TheAdversary has quit [Remote host closed the connection]
speeder39 has quit [Quit: Connection closed for inactivity]
ciphergoth has joined #pypy
ciphergoth has quit [Ping timeout: 250 seconds]
arigato has joined #pypy
<arigato>
seems I know the next thing to improve in CFFI
<arigato>
in CPython (particularly 3.x), removing "Py_BEGIN_ALLOW_THREADS" in the generated C code makes the program run 10x faster, in case there are multiple threads calling the C function all the time
<arigato>
just to be clear, here I'm complaining about the implementation of Py_BEGIN_ALLOW_THREADS
<arigato>
at least for this specific use case
domenkozar has quit [Ping timeout: 268 seconds]
ielectric has joined #pypy
realitix has joined #pypy
kai_w has joined #pypy
crane has quit [Read error: Connection reset by peer]
<kenaan>
arigo cffi/cffi 1443130c25e1 /c/_cffi_backend.c: Use memcpy here
<kenaan>
arigo default beeebc0b35f3 /pypy/module/_cffi_backend/: Test and fix: can get a segfault instead of an IndexError "too many initializers"
<kenaan>
arigo default 60409e95c501 /pypy/module/_cffi_backend/test/test_fastpath.py: more tests, trying to cover all paths
<kenaan>
arigo default feb6754915bc /: update to cffi/1443130c25e1
Hasimir has quit [Read error: Connection reset by peer]
TheAdversary has quit [Read error: Connection reset by peer]
Hasimir has joined #pypy
TheAdversary has joined #pypy
arigato has quit [Quit: Leaving]
TheAdversary has quit [Disconnected by services]
TheAdversary has joined #pypy
Khayman has joined #pypy
Hasimir has quit [Ping timeout: 240 seconds]
Khayman is now known as Hasimir
jamesaxl has quit [Ping timeout: 264 seconds]
jamesaxl has joined #pypy
<kenaan>
arigo cffi/cffi 960ab54c2eb4 /c/_cffi_backend.c: oops, don't do that if 'ct' is a CT_POINTER
marky1991 has joined #pypy
realitix has quit [Quit: realitix]
<kenaan>
arigo default 6df19fd2b6df /pypy/module/_cffi_backend/: update to cffi/960ab54c2eb4
dddddd has joined #pypy
mattip has joined #pypy
Khayman has joined #pypy
Khayman has quit [Changing host]
Khayman has joined #pypy
TheAdversary has quit [Disconnected by services]
TheAdversary has joined #pypy
Hasimir has quit [Ping timeout: 240 seconds]
Khayman is now known as Hasimir
fryguybob has quit [Ping timeout: 268 seconds]
ronan_ has joined #pypy
Rhy0lite has joined #pypy
<cfbolz>
mattip: hey, how did your talk go?
marky1991 has quit [Read error: Connection reset by peer]
<mattip>
cfbolz: The tutorial went ok, people on Mac OS and Windows had problems building the shared object that we wrapped in cffi. ctypes, cython, and cppyy
<mattip>
so now I have two issues on the github repo :)
marky1991 has joined #pypy
<cfbolz>
right, the usual problem if you have little time
<cfbolz>
mattip: but you also had a numpy talk this morning, right?
<mattip>
the numpy 15 minute overview went well
<cfbolz>
cool
<mattip>
it started some nice interactions with people