00:06
fotis has quit [Ping timeout: 250 seconds]
00:36
fotis has joined #pypy
01:20
fotis has quit [Ping timeout: 248 seconds]
01:47
fotis has joined #pypy
02:24
fotis has quit [Ping timeout: 240 seconds]
02:25
fotis has joined #pypy
02:31
fotis has quit [Ping timeout: 250 seconds]
02:33
fotis has joined #pypy
02:39
fotis has quit [Ping timeout: 250 seconds]
02:50
lritter has quit [Ping timeout: 252 seconds]
02:58
gcaldwell has joined #pypy
03:08
lritter has joined #pypy
03:14
fotis has joined #pypy
03:49
fotis has quit [Ping timeout: 245 seconds]
03:51
gcaldwell has quit [Ping timeout: 250 seconds]
04:25
fotis has joined #pypy
04:42
_whitelogger has joined #pypy
04:45
Corbin has quit [*.net *.split]
04:45
bbot2 has quit [*.net *.split]
04:45
samth has quit [*.net *.split]
04:45
energizer has quit [*.net *.split]
04:45
marvin_ has quit [*.net *.split]
04:45
Lightsword has quit [*.net *.split]
04:45
bbot2 has joined #pypy
04:45
marvin_ has joined #pypy
04:45
samth has joined #pypy
04:46
Lightsword has joined #pypy
04:47
energizer has joined #pypy
04:55
Corbin has joined #pypy
05:00
fotis has quit [Ping timeout: 250 seconds]
05:34
fotis has joined #pypy
05:35
<
mattip >
hey gcaldwell (for the logs)
05:36
<
mattip >
most of us are on european times, so if you have a question you should ask it, someone will eventually answer
05:39
lritter has quit [Ping timeout: 250 seconds]
06:09
fotis has quit [Ping timeout: 252 seconds]
06:36
glyph has quit [Quit: End of line.]
06:37
fotis has joined #pypy
06:37
glyph has joined #pypy
06:39
<
mattip >
the postgres-feedstock on conda-forge is failing to build with pypy, since postgres has its own datetime.h
06:39
<
mattip >
we include datetime.h in Python.h (CPython does not include it)
06:40
<
mattip >
we need it since we mangle the names of the datetime interfaces in pypy_decl.h, the name mangling fails if datetime.h is not included
06:43
<
mattip >
ideally we should move away from this name mangling, as suggested in
06:56
Atque has quit [Quit: ...]
07:10
greedom has joined #pypy
07:11
fotis has quit [Ping timeout: 240 seconds]
07:14
fotis has joined #pypy
07:23
<
mattip >
if I translate the jit-fewer-bridges branch locally and run gdb --args ./pypy-c pytest.py pypy/module/pypyjit/test_pypy_c
07:23
<
mattip >
I get a segfault in pypy_g_AbstractResumeGuardDescr_handle_fail
07:26
<
mattip >
that function (in rpython/jit/metainterp/compile.py) ends with an assert 0
07:26
<
mattip >
but I don't see any "return" statements that would exit the function before the assert 0
07:28
fotis has quit [Ping timeout: 252 seconds]
07:28
<
mattip >
I guess the other paths all run forever?
07:31
<
cfbolz >
mattip: hm, that's great I think
07:31
<
cfbolz >
it means we can find the segfault
07:31
<
cfbolz >
(it doesn't happen in any of the benchmarks)
07:31
<
cfbolz >
I'll look into it
07:41
mgorny has joined #pypy
08:00
fotis has joined #pypy
08:03
<
mattip >
I think it is escaping from self._trace_and_compile_from_bridge()
08:05
greedom has quit [Remote host closed the connection]
08:07
<
cfbolz >
mattip: yeah, that is supposed to always raise
08:30
<
mattip >
right, now I am starting to understand how this is meant to work
08:30
<
cfbolz >
mattip: it's all a bit indirect
08:31
* mattip
adding a "assert False" to the end of Metainterp.handle_guard_failure()
08:34
fotis has quit [Ping timeout: 244 seconds]
08:36
fotis has joined #pypy
08:40
fotis has quit [Ping timeout: 240 seconds]
09:11
fotis has joined #pypy
09:26
fotis has quit [Ping timeout: 250 seconds]
09:50
<
cfbolz >
mattip: which of the test_pypy_c tests is failing?
10:07
<
cfbolz >
mattip: ah, wait
10:08
<
cfbolz >
mattip: we should merge open-ended-traces to py3.8, not jit-fewer-bridges
10:08
<
cfbolz >
the latter is very much in progress still
10:54
fotis has joined #pypy
10:58
fotis has quit [Ping timeout: 240 seconds]
10:59
stkrdknmibalz has quit [Quit: WeeChat 3.0.1]
11:47
wisefriday has joined #pypy
12:54
fotis has joined #pypy
12:59
fotis has quit [Ping timeout: 244 seconds]
14:14
gef_ has quit [Read error: Connection reset by peer]
14:17
gef_ has joined #pypy
14:31
fotis has joined #pypy
14:35
fotis has quit [Ping timeout: 245 seconds]
15:39
vstinner has joined #pypy
15:40
<
vstinner >
i'm talking about the Python C API using "kind" and "data" which are concepts very specific to the exact PEP 393 implementation
15:41
<
vstinner >
i know that PyPy uses UTF-8, whereas CPython uses UCS-1, UCS-2 or UCS-4 depending on the string content. Inada-san considers to switch CPython to UTF-8, but I don't know if he tried to implement his idea
16:14
gcaldwell has joined #pypy
16:32
fotis has joined #pypy
16:36
fotis has quit [Ping timeout: 244 seconds]
16:54
<
mattip >
vstinner: I recently implemented compact strings for PyPy, following the CPython implementation
16:58
<
mattip >
we try to respect the KIND PyUnicode API
16:59
<
mattip >
actually, Inada-san could take what we do, fix the errors, and use it
17:37
gef_ has quit [Ping timeout: 256 seconds]
17:38
gef has quit [Ping timeout: 240 seconds]
17:38
gef has joined #pypy
17:44
gef_ has joined #pypy
17:59
gcaldwell has quit [Remote host closed the connection]
18:32
fotis has joined #pypy
18:36
fotis has quit [Ping timeout: 240 seconds]
19:00
fotis has joined #pypy
19:04
fotis has quit [Ping timeout: 252 seconds]
21:00
fotis has joined #pypy
21:04
fotis has quit [Ping timeout: 250 seconds]
21:21
gef has quit [Ping timeout: 252 seconds]
21:21
gef_ has quit [Ping timeout: 250 seconds]
21:22
fotis has joined #pypy
21:26
fotis has quit [Ping timeout: 250 seconds]
21:29
gef has joined #pypy
21:33
gef_ has joined #pypy
21:41
gef has quit [Ping timeout: 244 seconds]
21:41
gef_ has quit [Ping timeout: 250 seconds]
22:12
gef has joined #pypy
22:12
gef_ has joined #pypy
22:16
gef has quit [Ping timeout: 250 seconds]
22:17
gef_ has quit [Ping timeout: 245 seconds]
22:18
gef has joined #pypy
22:21
gef_ has joined #pypy
22:25
gef has quit [Ping timeout: 252 seconds]
22:25
gef_ has quit [Ping timeout: 256 seconds]
22:33
gef has joined #pypy
22:35
gef_ has joined #pypy
22:45
Atque has joined #pypy
23:11
stkrdknmibalz has joined #pypy