These are chat archives for dropbox/pyston

23rd
Sep 2015
Kevin Modzelewski
@kmod
Sep 23 2015 00:48
hmm @daetalus I'm seeing some test failures in test_complex.py if I update libpypa to include your latest change
An Long
@aisk
Sep 23 2015 04:53
@Daetalus @undingen thx~
Sun
@Daetalus
Sep 23 2015 06:12
@kmod I wil take a look
An Long
@aisk
Sep 23 2015 08:20
got this
pyston $ make [2] ninja -C /home/vagrant/pyston/build/Debug pyston copy_stdlib copy_libpyston sharedmods ext_pyston ext_cpython ninja: Entering directory `/home/vagrant/pyston/build/Debug' ninja: error: '../../from_cpython/Parser/asdl.py', needed by 'from_cpython/Parser/asdl.py', missing and no known rule to make it make: *** [pyston_dbg] Error 1
while running make in the master branch 😂
Ignore me .... solved it by removed the build folder. I think my local ninja file not updated.
Kevin Modzelewski
@kmod
Sep 23 2015 09:58
yeah, we use some filename wildcards inside cmake, and it caches the evaluation of those
so you have to touch CMakeLists.txt to get that to rerun
Sun
@Daetalus
Sep 23 2015 10:00
Hi, @kmod , I will use CPython logic in setContain. But would you mind to explain why current implementation different than CPython a bit? I subclass the set and add a __hash__ that throw exception. Nothing different.
Kevin Modzelewski
@kmod
Sep 23 2015 10:26
This is silly, but this throws an exception in cpython and not for us:
class MySet(set):
    def __hash__(self):
        print "calling __hash__"
        raise ValueError()

print MySet() in set()
your code also uses a slightly different lookup rule ("try __eq__ against any other sets in the set"), where cpython's is "look up the hash table slot and then call __eq__"
Sun
@Daetalus
Sep 23 2015 10:56
Thanks!