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
<tos9>
does pypy have different default include directories than cpython (within cffi?)
<tos9>
pypy seems to happily find libraries in /usr/local, but CPython doesn't for me
<mattip>
some of this stuff comes from distutils, which uses sysconfig.get_config_vars(x) to determine compilation flags
<mattip>
so check what CPython and PyPy give you for x in ('CFLAGS', 'CPPFLAGS', 'CONFIGURE_CPPFLAGS', 'CFLAGSFORSHARED') or maybe others
<tos9>
thanks, was just perusing `python -m sysconfig` and didn't see any difference
* tos9
checks those instead
dddddd has joined #pypy
<tos9>
Hrm, they're all None on PyPy (which works) and all not None on CPython which doesn't
<tos9>
but neither mention /usr/local, CPython here (from Homebrew) sets CFLAGS to include just a bunch of directories from XCode
<tos9>
ah, maybe they're using different compilers entirely
<mattip>
you should be able to set verbose=True somewhere and get the build output
<tos9>
ah cool thanks
<tos9>
and indeed, pypy is using gcc and cpython clang... ok
<tos9>
was driving me nuts seeing how many builds in GHA had different behavior in pypy vs cpython for reasons that shouldn't make sense
<tos9>
bleh, that doesn't explain it though, because gcc *is* clang on osx right, ok, it is then one of the flags... /me tries removing them one by one
<mattip>
finding libraries or headers? linker or compiler? maybe it is a different set of config_vars
<tos9>
it's failing first finding headers
<tos9>
but the linker may fail too if it gets passed that
<tos9>
I managed to break pypy now by exporting CFLAGS to a similar value
<tos9>
OK apparently it's -isysroot that's being set
<tos9>
ok well since I got tired of trying to figure out how to get CPython to *not* set isysroot or whatever I just set c_include_path=(/usr/local/include) and everything now compiles on both... which helped me find my segfault (that I was calling ffi.gc but returning the original thing, I think I've made that mistake 30 times already)
_whitelogger has joined #pypy
tsaka__ has quit [Ping timeout: 260 seconds]
adamholmberg has joined #pypy
<cfbolz>
mattip: I got the rsplit bug, fix coming soon