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)"
igitoor has quit [Changing host]
igitoor has joined #pypy
Hotpot33 has quit [Ping timeout: 248 seconds]
astronavt has quit [Quit: Leaving...]
astronavt has joined #pypy
tbodt has quit [Ping timeout: 248 seconds]
tbodt has joined #pypy
webmeister has joined #pypy
tormoz has joined #pypy
tbodt has quit [Ping timeout: 268 seconds]
adamholmberg has quit [Remote host closed the connection]
<arigato>
glyph: accessing fields of PyDateTime objects was never supposed to work
<arigato>
I guess it's not a regression, but a #ifdef inside psycopg2
amaury has joined #pypy
zmt00 has quit [Quit: Leaving]
lritter has quit [Remote host closed the connection]
ionelmc has joined #pypy
<glyph>
arigato: so the bug is in psycopg2?
<fijal>
arigato: the bug is something stupid, like we have a rsre_utf8.Utf8MatchContext but ctx.w_unicode is NULL
<fijal>
arigato: yes, fresh_copy does not make a copy of w_unicode_obj
<fijal>
how do I test it?
<njs>
glyph: it looks like the official way to get at the innards of a PyDateTime object (in CPython) is macro accessors like PyDateTime_GET_MONTH, ..._GET_DAY, etc.
<arigato>
fijal: there is another bug?
<fijal>
arigato: did you fix one yesterday?
<arigato>
yes
<arigato>
I sometimes follow up when I promize I'll fix one bug :-)
<arigato>
it's probably a different one, but yesterday all of cpython's test_re passed
<fijal>
Gah didn't look apparently
<fijal>
Ok, fair.
<fijal>
Tell me who copies w_unicode_obj when fresh ctx is made when calling fresh copy?
<arigato>
it's probably easy to test, I think that findall() (?) uses fresh_copy
<arigato>
well, just look
<arigato>
or I can do it if you prefer
<arigato>
(the bug is clearly that I forgot to copy w_unicode_obj)
<fijal>
Ok
<fijal>
I'll look
<arigato>
I guess fresh_copy should be removed from rsre/, it's only called from pypy/module/_sre anyway
<fijal>
But I also lied to myself about whether I looked if you commited something apparently
<fijal>
Ok
<fijal>
so I would not trust myself ;-)
exarkun has quit [Read error: Connection reset by peer]
<fijal>
arigato: can I kill todays benchmark runs? or do I have to wait till 4pm before mine starts?
<arigato>
sure I guess
<fijal>
it's a bit bad, I agree
<arigato>
ah, I need to look around anyway, there's another problem here
<fijal>
arigato: can you kick the benchmark buildbot once you're done please?
<arigato>
yes
<kenaan>
arigo unicode-utf8 d978136e55f3 /: Tests and fixes * cannot access a position greater than ctx.end; need some small refactorings and added an ...
<arigato>
trying to translate and run cpython's test_re, first
gclawes has quit [Read error: Connection reset by peer]
gclawes has joined #pypy
<kenaan>
cfbolz default a00ca61351ba /: merge rdict-fast-hash: make it possible to declare that the hash and eq functions used in an objectmodel.r_dict a...
<kenaan>
rlamy unicode-utf8 9450e5936f78 /lib-python/2.7/test/test_memoryio.py: Mark test CPython-only - the test fails with _pyio, even on CPython
antocuni__ has quit [Ping timeout: 248 seconds]
ceridwen has quit [Ping timeout: 255 seconds]
<cfbolz>
pff, translation is broken on the branch :-(
<cfbolz>
fijal, arigato: does append_code take an unsigned int or a signed int?
<cfbolz>
on the utf8builder
<arigato>
your guess is as good as mine at this point, but I think we should be phasing out r_uint from these interfaces
marr has joined #pypy
<kenaan>
amauryfa py3.6 07c74faf3629 /pypy/module/posix/: os.fspath() accepts a DirEntry
<kenaan>
amauryfa py3.6 10f89fd70107 /pypy/module/: Most file operations now accept a PathLike object with a __fspath__ method.
<tos9>
antocuni (for logs in case you read them, otherwise will find you tomorrow): so, finally getting back to smartstrip stuff -- I see though first that one of the current tests actually seems to fail for me on OSX
<tos9>
specifically test_strip, which seems odd, because well, that's the one that tests that stuff was stripped, which seems like it's working properly given that the binaries are actually stripped :)