00:00
<
marky1991 >
oh, pretty late in europe. I'll ask again tomorrow
00:07
oberstet has quit [Quit: Leaving]
01:08
marky1991 has quit [Quit: Leaving]
01:25
lritter has quit [Ping timeout: 256 seconds]
01:39
mjacob has quit [Ping timeout: 246 seconds]
02:33
mjacob has joined #pypy
03:14
Taggnostr has quit [Remote host closed the connection]
03:15
Taggnostr has joined #pypy
03:30
mgedmin has joined #pypy
03:33
jcea has quit [Ping timeout: 264 seconds]
06:20
isidentical has joined #pypy
06:53
<
mattip >
marky1991: that clearly is wrong. Want to submit a PR to fix it?
07:46
krono has joined #pypy
09:10
oberstet has joined #pypy
09:11
oberstet has quit [Remote host closed the connection]
09:11
oberstet has joined #pypy
09:39
<
cfbolz >
marky1991: yeah, that looks wrong somehow, // would make more sense
10:24
glyph has quit [Quit: End of line.]
10:25
glyph has joined #pypy
11:26
marky1991 has joined #pypy
11:27
<
marky1991 >
ok, i'll probably work on a fix then. I now see why the +- logic is there, need to run through some examples and add some tests
11:27
<
marky1991 >
i"m surprised no one ever hit it though, the code in ctypes is 10+ years old
11:28
<
marky1991 >
i'm hitting it due to pyglet
11:28
<
marky1991 >
but thanks for confirming
11:35
<
cfbolz >
marky1991: because in PyPy2 it's fine, I suspect
11:36
<
cfbolz >
(there, / has the behavior of // by default)
11:36
<
marky1991 >
ah, right
11:37
<
marky1991 >
not a lot of existing tests it seems?
11:37
<
marky1991 >
i see stdlib test_ctypes, which is pretty bare bones and then extra_tests/test_ctypes, which also isn't very big
11:37
<
marky1991 >
should I add the new tests to extra_tests?
11:38
<
marky1991 >
i've never used that folder before
11:39
<
marky1991 >
oh, it looks like there's a .c test as well, hmm
11:41
<
mattip >
marky1991: yes, extra_tests is the right place
12:35
jcea has joined #pypy
13:18
cptpcrd2 has left #pypy ["WeeChat 2.3"]
13:42
marky1991 has quit [Ping timeout: 256 seconds]
13:45
marky1991 has joined #pypy
14:17
marky1991 has quit [Ping timeout: 260 seconds]
15:14
marky1991 has joined #pypy
15:15
marky1991 has quit [Remote host closed the connection]
15:15
marky1991 has joined #pypy
15:24
<
marky1991 >
im struggling to build pypy3 , due to a ssl compilation error
15:25
<
marky1991 >
hmm, maybe I don't exactly care, let me try something
15:43
marky1991 has quit [Ping timeout: 240 seconds]
15:58
asmeurer has joined #pypy
16:29
lritter has joined #pypy
16:37
<
mattip >
marky1991 (for the logs): you can fix the cytpes issue in pure python, no need to translate pypy
16:49
asmeurer has quit [Quit: asmeurer]
17:19
<
fijal >
eh, strange question
17:19
<
fijal >
would it make sense to distribute "C sources" from compilation as a way to address linux binary compatibility problems
17:28
<
mattip >
what is the use case?
17:29
<
mattip >
btw, currently building py3.7 takes ~30 minutes, which is comparable to building pytorch/tensorflow
17:29
<
cfbolz >
on a sufficiently fast machine ;-)
17:30
<
mattip >
we could easily add another tarball of C sources to the release
17:30
<
mattip >
just wondering who is the target audience
17:34
<
fijal >
mattip: different version of ubuntu is target audience
17:34
<
fijal >
so right now what I need to do is to fix the docker images, because the version of libffi changed
17:42
<
mattip >
the releases were made to be portable
17:42
<
mattip >
between glibc linuxes
17:47
<
mattip >
and ships libffi.so.6
17:55
<
fijal >
right, so if it's libffi.so.7, boom
17:57
isidentical has quit [Read error: Connection reset by peer]
17:57
isidentical1 has joined #pypy
17:59
<
cfbolz >
fijal:
*ships*
18:00
oberstet has quit [Quit: Leaving]
18:00
isidentical1 is now known as isidentical
18:02
max3 has joined #pypy
18:02
<
max3 >
i can't find the src anywhere
18:03
<
cfbolz >
it's still a branch, I suppose
18:04
<
max3 >
cfbolz thanks
18:04
max3 has quit [Client Quit]
18:04
marky1991 has joined #pypy
18:04
<
marky1991 >
could someone remind me how to run the tests untranslated?
18:05
<
marky1991 >
i tried just python ../../pytest.py test_array.py , but that's just testing cpython
18:10
<
cfbolz >
marky1991: sorry, we're not explaining this very well. do you have a translated pypy around?
18:11
<
cfbolz >
if yes, you can edit the ctypes sources, and run the test on the
*translated* pypy
18:11
<
cfbolz >
by saying pypy-c -m pytest ...
18:12
<
fijal >
cfbolz: why does it not find it then?
18:12
<
cfbolz >
I am not sure this stuff is integrated into the arm build
18:13
<
fijal >
ok, how do I integrate it into the arm build?
18:13
<
fijal >
or, specifically, what's "this stuff"
18:13
<
cfbolz >
fijal: squeaky's portable builds are not in our repo
18:14
<
fijal >
so you are talking about portable builds and not about nightlies?
18:14
<
cfbolz >
I am just adding to the confusion, I should eat
18:15
<
fijal >
ok, cool, I can put some work into making it happen for arm64 too
18:21
<
marky1991 >
ok, i got it with -m pytest as suggested
18:22
<
cfbolz >
marky1991: it's all written in pure python
18:22
<
cfbolz >
so it's relatively regular library development
18:22
<
marky1991 >
just difficult to get it to use the right python
18:30
dmalcolm has quit [Remote host closed the connection]
20:04
marky1991 has quit [Ping timeout: 240 seconds]
21:15
mattip has quit [Ping timeout: 265 seconds]
21:23
mattip has joined #pypy
22:10
ronan has quit [Remote host closed the connection]
22:10
ronan has joined #pypy
22:11
ronan is now known as Guest59203
22:22
Guest59203 is now known as ronan
22:24
mattip has quit [Ping timeout: 246 seconds]
22:33
mattip has joined #pypy
22:49
isidentical has quit [Quit: isidentical]
23:38
<
exarkun >
maybe my pypy is out of date, but it seems pypy2.7 os.execvp uses the default encoding instead of the filesystem encoding when it gets unicode arguments
23:38
<
exarkun >
cpython uses filesystem encoding