antocuni changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | "PyPy: the Gradual Reduction of Magic (tm)"
<kenaan_>
cfbolz default 57349ce40416 /pypy/interpreter/pyframe.py: no need to get the executioncontext a second time, it's already in a local variable
<kenaan_>
cfbolz call-loopinvariant-into-bridges 7d193b453f44 /rpython/jit/metainterp/test/test_bridgeopt.py: a branch to try to emit less threadlocalref_get calls in bridges first failing test
raynold has quit [Quit: Connection closed for inactivity]
ssbr` has quit [Ping timeout: 265 seconds]
<_kcr>
So I found a place where the pypy3 _curses module doesn't behave like the cpython one. I have a fix. What's the procedure to submit it these days? (it's not big)
<kenaan>
arigo default 89245c285fa0 /pypy/module/_sre/: Issue #2758 Test and fix for re.sub(callback) receiving an immutable match object
mattip has joined #pypy
<_kcr>
arigato: thanks
forgottenone has joined #pypy
forgottenone has quit [Quit: Konversation terminated!]
forgottenone has joined #pypy
oberstet has joined #pypy
forgottenone has quit [Quit: Konversation terminated!]
slacky has joined #pypy
oberstet has quit [Ping timeout: 256 seconds]
<nanonyme>
Hmm, interestingly looks like CFFI is starting to natively support a lot of the things I wrote simplecpreprocessor to begin with. Not #include in ABI mode though (yet?)
altendky has quit [Quit: Connection closed for inactivity]
<mattip>
nanonyme: #include will IMO be a tough sell, it is tied to all kinds of magic
asmeurer has joined #pypy
asmeurer has quit [Client Quit]
<nanonyme>
mattip, well, once you have defines, ifdefs and such, include is mostly a matter of a matter of flattening multiple header files into on header file *shrug*
<nanonyme>
That said, I did have to special-case a couple of headers (eg windows.h) so its include is just silently ignored
<mattip>
AFAIK we don't have ifdefs
<mattip>
and the issue with include is _where_ exactly you find those multiple headers
<mattip>
sometimes it is non-trivial to figure out, like windows SDK versions, looking stuff up in the registry, strange environment variables, ...
<nanonyme>
Yeah, well, for that reason simplecpreprocessor allows injecting definitions by the consumer of the API
<nanonyme>
During actual traversing no environment variables are read and nothing is checked from registry
<nanonyme>
(although I'm to be honest going to start promoting using bootstrapper headers so you define macros in your bootstrapper header, then include the actual header)
energizer has quit [Remote host closed the connection]
energizer has joined #pypy
adamholmberg has joined #pypy
adamholmberg has quit [Ping timeout: 256 seconds]
antocuni has quit [Ping timeout: 256 seconds]
oberstet has quit [Ping timeout: 240 seconds]
<mattip>
microsoft has updated the vs 2017 compiler to 1500, and apparently broke distutils/setuptools again
<mattip>
so it is really painful to try to build numpy for the newer vs2017 builds of pypy3.5
<mattip>
I can't even get to a point where the vsvarsall runs properly after setting VS140COMNTOOLS
slacky has quit [Quit: Leaving]
<nanonyme>
Well, MS did say the environment variables will not be used again because the entire idea is that one machine can have multiple VS 2017 installations
<nanonyme>
Instead you're supposed to use an executable which digs and outputs paths to VS 2017 installations
<mattip>
do you have a link to that info?
<mattip>
or the name of the exe?
adamholmberg has joined #pypy
<mattip>
AFAICT even on cpython HEAD, msvc9compiler will still respect VS140COMNTOOLS
adamholmberg has quit [Ping timeout: 240 seconds]
asmeurer_ has joined #pypy
altendky has joined #pypy
mattip has left #pypy ["Leaving"]
fryguybo1 has quit [Ping timeout: 256 seconds]
fryguybob has joined #pypy
squeaky_pl has quit [Ping timeout: 256 seconds]
njs has quit [Ping timeout: 256 seconds]
njs has joined #pypy
drolando_ has quit [Remote host closed the connection]