cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://botbot.me/freenode/pypy/ ) | use cffi for calling C | "the modern world where network packets and compiler optimizations are effectively hostile"
<exarkun>
I have a patch to buildbot to have it automatically clean up hg bookmarks before each run. Maybe. Untested.
rokujyouhitoma has quit [Ping timeout: 252 seconds]
<exarkun>
someone (maybe ronny?) suggested deleting .hg/bookmarks in the slave checkout manually might also help (that's all my buildbot changes do, but automatically)
* ronny
steped down from anything hg related way before
<fijal>
haha :)
<exarkun>
ah sorry, it was mattip
<ronny>
reason: atlassian made bitbucket total pain
<exarkun>
fijal: I guess at least some parts of test_cmd_line_script look tractable
<exarkun>
File "/tmp/test.py", line 17, in <module>
<exarkun>
from twisted.trial.unittest import TestCase
<exarkun>
ImportError: No module named 'twisted'
<exarkun>
looks like an exciting one.
<fijal>
exarkun: it looks like they're trying to write /tmp/test.py and run it
<fijal>
(an EXCELLENT idea)
<exarkun>
what could possibly go wrong.
<fijal>
maybe not?
<exarkun>
anyway, I'll take a look at it when I next get a chance
<exarkun>
How do I make a new hg branch? I can push my buildbot changes so other folks can see them, at least.
<fijal>
hg branch new-branch
<fijal>
hg push --new-branch
<fijal>
exarkun: what buildslave/hg problems? do you have a link?
<exarkun>
I don't think there are any primary artifacts that reflect the problem. Or, none I know of, at least. The
<fijal>
exarkun: armin says he's not aware that the problems are there any more (he did stuff)
<exarkun>
a week or two ago I accidentally put a py3.5 bookmark onto default. I thought I cleaned it up. Then arigo thought he cleaned it up. But mattip said it was still messing things up on the build slave.
<fijal>
ok
<fijal>
mattip: ^^^
<exarkun>
I just looked at the slave on bencher4
<exarkun>
/extra1/home/buildslave/buildslave/own-linux-x86-64/build/.hg/bookmarks has a line in it for py3.5
<exarkun>
I guess that's probably wrong
<exarkun>
dunno why it is still there and dunno why other builders' checkouts don't (seem to) also have that problem
<exarkun>
I guess deleting that file will fix it though
<kenaan>
arigo py3.5 b4d19ffb3525 /lib_pypy/pyrepl/readline.py: Issue #2638 Workaround for .python_history files that have non-utf-8 chars. But ideally we should encode/decode the...
<cfbolz>
I must say that is a bizarre python-dev idea
<arigato>
cfbolz: yup
<kenaan>
arigo py3.5 c1fb69da92dc /pypy/objspace/std/: Test and fix (lib-python/3/test/test_descr)
ronny has quit [Ping timeout: 252 seconds]
raynold has joined #pypy
rokujyouhitoma has joined #pypy
<kenaan>
arigo py3.5 e8a566e9fcd9 /pypy/objspace/std/: Refinement of c1fb69da92dc
<fijal>
cfbolz: now he's debugging "what happens if I add new subclasses inside overwritten __mro__"
<cfbolz>
I bet that there is a CPython crash in that area
<fijal>
yes that's what armin said :)
rokujyouhitoma has quit [Ping timeout: 252 seconds]
<exarkun>
such useful features
<cfbolz>
I'm sure
oberstet has quit [Ping timeout: 240 seconds]
<arigato>
failed to find the CPython crasher
<fijal>
impressive
<kenaan>
arigo py3.5 c30f2d4e721f /pypy/objspace/std/test/test_typeobject.py: This case "works", but gives a slightly strange error message on both CPython and PyPy
asmeurer has quit [Quit: asmeurer]
bgola has quit [Quit: Lost terminal]
<kenaan>
arigo default 18b10a6743c4 /: (fijal, arigo) Rename _pypy_dll to __pypy_dll__, so that the custom __getattr__ would raise if the attribute doesn...
<kenaan>
arigo default b02f6ce0d05b /pypy/module/test_lib_pypy/README.txt: (fijal, arigo) Improve the explanation a bit
mvantellingen has joined #pypy
oberstet has joined #pypy
<kenaan>
cfbolz regalloc-playground c02b584c919d /rpython/jit/backend/llsupport/test/test_regalloc.py: now the coalescing integration test works
<kenaan>
cfbolz regalloc-playground 3ba71498fe77 /rpython/jit/backend/llsupport/test/test_regalloc.py: more realistic jump faking
<kenaan>
cfbolz regalloc-playground 3cde3ad8b828 /rpython/jit/backend/llsupport/test/test_regalloc.py: make it possible to specify the locs of the inputargs in the tests. a skipped test that's messy to fix.
<kenaan>
cfbolz regalloc-playground f68f729a80cf /rpython/jit/backend/llsupport/: use the regular spill heuristics to chose the variables to spill before a call
tbodt has joined #pypy
<kenaan>
cfbolz regalloc-playground ac35a22e47df /: just to check that jump hinting would work too
<fijal>
mattip: any reason why PyType_IS_GC returns 0?
<fijal>
(it returns 1 on py3k and we're debating it)
<mattip>
no particular reason, cython does things like "if PyType_IS_GC(o) PyObject_GC_Track(o)"
<mattip>
but since PyObject_GC_Track is a NOOP anyway I thought to save the function call
<arigato>
ok, so we can #define PyObject_GC_Track to do nothing anyway