00:00
kipras`away is now known as kipras
00:05
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
00:23
aboudreault has quit [Excess Flood]
00:27
GothAlice is now known as PresidentCthulhu
00:27
PresidentCthulhu is now known as GothAlice
00:27
aboudreault has joined #pypy
00:30
yuyichao_ has quit [Ping timeout: 240 seconds]
00:40
marr has quit [Ping timeout: 246 seconds]
00:56
tilgovi has joined #pypy
00:57
tbodt has joined #pypy
00:57
yuyichao_ has joined #pypy
01:00
mat^2 has quit [Quit: Leaving]
01:13
jcea has quit [Quit: jcea]
01:22
tbodt has quit [Ping timeout: 260 seconds]
01:29
o11c|vacation is now known as o11c
01:41
tbodt has joined #pypy
02:09
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
02:15
wsanchez has joined #pypy
02:17
ArneBab has joined #pypy
02:21
ArneBab_ has quit [Ping timeout: 240 seconds]
03:22
pilne has quit [Quit: Quitting!]
03:50
egregius313 has joined #pypy
03:56
yuyichao_ has quit [Quit: Konversation terminated!]
03:57
yuyichao has joined #pypy
04:11
egregius313 has quit [Remote host closed the connection]
04:12
egregius313 has joined #pypy
04:16
egregius313 has quit [Ping timeout: 246 seconds]
04:34
fotis has quit [Quit: No Ping reply in 180 seconds.]
04:34
fotis2 has quit [Quit: No Ping reply in 180 seconds.]
04:36
fotis has joined #pypy
04:37
fotis2 has joined #pypy
04:53
_whitelogger has joined #pypy
05:00
fotis has quit [Quit: No Ping reply in 180 seconds.]
05:01
fotis2 has quit [Quit: No Ping reply in 180 seconds.]
05:01
fotis has joined #pypy
05:02
fotis2 has joined #pypy
05:10
jamadden has quit [Quit: Leaving.]
05:10
egregius313 has joined #pypy
05:12
jacob22_ has quit [Quit: Konversation terminated!]
05:41
egregius313 has quit [Remote host closed the connection]
05:41
egregius313 has joined #pypy
05:45
<
Alex_Gaynor >
tumbleweed:
05:45
jamesaxl has joined #pypy
05:45
egregius313 has quit [Ping timeout: 246 seconds]
05:51
jamadden has joined #pypy
05:51
jamadden has quit [Client Quit]
06:00
jamadden has joined #pypy
06:00
jamadden has quit [Client Quit]
06:50
jamescampbell has joined #pypy
06:51
amaury has joined #pypy
07:18
oberstet has joined #pypy
07:20
jamescampbell has quit [Quit: Leaving...]
07:21
wsanchez has left #pypy [#pypy]
07:22
realitix has joined #pypy
07:50
jamescampbell has joined #pypy
07:51
arigato has joined #pypy
07:53
Joannah has joined #pypy
08:12
marr has joined #pypy
08:15
jamescampbell has quit [Remote host closed the connection]
08:16
jamadden has joined #pypy
08:16
jamadden has quit [Client Quit]
08:36
antocuni has joined #pypy
08:44
Joannah has quit [Ping timeout: 260 seconds]
08:50
cstratak has joined #pypy
08:55
<
antocuni >
arigato: did you have a chance to review my branch, btw?
08:56
<
arigato >
no, sorry
09:09
<
arigato >
antocuni: is there a way to complain more strongly if we forget @GCBuffer.decorate?
09:10
<
antocuni >
arigato: I fear the only way is to use a metaclass
09:10
<
antocuni >
which I preferred not to use, to avoid too much magic
09:10
<
arigato >
would "NOT_RPYTHON" in the default typed_read() and typed_write() work?
09:11
<
arigato >
ah no, not if it's called generically
09:12
<
antocuni >
I suppose we could have a flag on GCBuffer which is overridden by @decorate, and the
__init__ which checks it
09:12
<
antocuni >
but it looks a bit obscure
09:12
<
arigato >
or more obscure playings with @specialize
09:12
<
antocuni >
I tried to play with @specialize by failed miserably
09:13
<
antocuni >
arigato: note that the only reason to inherit from GCBuffer is to get the default implementations of typed_read and typed_write
09:13
<
arigato >
the alternative is to use a mixin
09:13
<
antocuni >
so your tests will immediately catch it
09:14
<
arigato >
just in case: I thought that you could write "class Foo(Buffer, GCBufferMixin)"
09:14
<
arigato >
which basically makes a copy the methods from GCBufferMixin, for the annotator
09:14
<
antocuni >
I
*think* I have tried that route as well
09:14
<
arigato >
import_from_mixin()
09:14
<
antocuni >
but it does not work because the base_ofs is not constant
09:15
<
antocuni >
here the trick is to call _get_gc_data_offset() outside the functions
09:15
<
arigato >
then, do you really need GCBuffer instead of only calling a class decorator?
09:16
<
antocuni >
I introduced it mostly to get a default implementation for _get_gc_data_extra_offset
09:17
<
arigato >
the class decorator can insert this function into the class, but only if it's not already defined
09:17
<
arigato >
well, it looks more magic again
09:17
<
antocuni >
yes I know and I even started to do that, but the implementation started to look a bit complex
09:18
<
antocuni >
arigato: 7538ecbcf8b5 is the checkin which turned GCBuffer from being a decorator to a base class
09:31
jacob22_ has joined #pypy
09:37
girish946 has joined #pypy
09:46
girish946 has quit [Quit: Leaving]
09:58
nimaje1 has joined #pypy
09:58
nimaje has quit [Killed (cherryh.freenode.net (Nickname regained by services))]
09:58
nimaje1 is now known as nimaje
10:13
antocuni has quit [Ping timeout: 246 seconds]
10:35
fniephaus has quit [Ping timeout: 240 seconds]
10:35
fniephaus has joined #pypy
10:35
amaury has quit [Ping timeout: 268 seconds]
10:46
amaury has joined #pypy
10:58
jamadden has joined #pypy
11:09
inad922 has joined #pypy
11:11
amaury has quit [Ping timeout: 240 seconds]
11:13
jcea has joined #pypy
11:16
cstratak_ has joined #pypy
11:16
cstratak has quit [Read error: Connection reset by peer]
11:18
amaury has joined #pypy
11:26
bremner has quit [Ping timeout: 246 seconds]
11:27
bremner has joined #pypy
11:27
bremner has quit [Changing host]
11:27
bremner has joined #pypy
11:40
bremner has quit [Ping timeout: 255 seconds]
11:41
Tiberium has joined #pypy
11:47
tilgovi has quit [Ping timeout: 246 seconds]
11:47
jacob22_ has quit [Ping timeout: 255 seconds]
11:48
arigato has quit [Quit: Leaving]
11:48
Knio_ has quit [Ping timeout: 240 seconds]
11:49
Knio has joined #pypy
11:49
fniephaus has quit [Ping timeout: 272 seconds]
11:49
bremner has joined #pypy
11:50
bremner has quit [Changing host]
11:50
bremner has joined #pypy
11:51
fniephaus has joined #pypy
11:58
kolko has quit [Ping timeout: 240 seconds]
12:02
<
kenaan >
Kounavi Kounavi/fix-typo-depricate-to-deprecate-p-1495624547235 9fe360fea796 /pypy/doc/windows.rst: Fix typo (depricate to deprecate) :P
12:02
<
kenaan >
arigo default 62e05f75166d /pypy/doc/windows.rst: Merged in Kounavi/pypy/Kounavi/fix-typo-depricate-to-deprecate-p-1495624547235 (pull request #550) Fix typo (depri...
12:04
arigato has joined #pypy
12:07
amaury has quit [Ping timeout: 245 seconds]
12:17
inad922 has quit [Ping timeout: 255 seconds]
12:20
amaury has joined #pypy
12:23
antocuni has joined #pypy
12:37
Tiberium has quit [Remote host closed the connection]
12:40
Tiberium has joined #pypy
12:53
amaury has quit [Ping timeout: 240 seconds]
13:15
amaury has joined #pypy
13:21
girish946 has joined #pypy
13:22
Tiberium has quit [Remote host closed the connection]
13:28
Tiberium has joined #pypy
13:32
realitix has quit [Quit: Leaving]
13:39
lritter_ has joined #pypy
13:43
<
cfbolz >
arigato: the pgo pull request looks good to me now. ok if I merge it, or do you want to take a look?
13:46
realitix has joined #pypy
13:49
Tiberium has quit [Remote host closed the connection]
13:56
realitix has quit [Ping timeout: 240 seconds]
13:58
Tiberium has joined #pypy
13:59
Remi_M has quit [Quit: See you!]
14:04
<
antocuni >
good, the win32 tests finished and faster-rstruct-2 is not more broken than default :)
14:05
jamescampbell has joined #pypy
14:06
<
cfbolz >
antocuni: in my opinion you can merge
14:06
<
antocuni >
arigato: do you want to do any more review or it's ok to you too?
14:10
realitix has joined #pypy
14:10
jcea has quit [Ping timeout: 260 seconds]
14:11
jcea1 has joined #pypy
14:13
jcea1 is now known as jcea
14:13
Tiberium has quit [Remote host closed the connection]
14:14
Tiberium has joined #pypy
14:17
jamescam_ has joined #pypy
14:18
jamescampbell has quit [Ping timeout: 246 seconds]
14:26
jamescampbell has joined #pypy
14:26
tbodt has joined #pypy
14:28
jamescam_ has quit [Ping timeout: 260 seconds]
14:29
jamescam_ has joined #pypy
14:32
jamescampbell has quit [Ping timeout: 268 seconds]
14:33
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
14:33
tbodt has joined #pypy
14:36
tbodt has quit [Client Quit]
14:36
jamescampbell has joined #pypy
14:38
jamesca__ has joined #pypy
14:40
jamescam_ has quit [Ping timeout: 246 seconds]
14:40
egregius313 has joined #pypy
14:41
jamescampbell has quit [Ping timeout: 246 seconds]
14:42
egregius313 has quit [Remote host closed the connection]
14:42
egregius313 has joined #pypy
14:44
<
arigato >
antocuni & cfbolz: you can merge
14:45
jamesca__ has quit [Read error: Connection reset by peer]
14:46
jcea has quit [Ping timeout: 268 seconds]
14:46
egregius313 has quit [Remote host closed the connection]
14:46
egregius313 has joined #pypy
14:51
girish946 has quit [Quit: Leaving]
14:51
jamescampbell has joined #pypy
14:53
<
antocuni >
arigato: ok thanks :)
14:54
realitix has quit [Ping timeout: 240 seconds]
14:55
jamescampbell has quit [Read error: Connection reset by peer]
14:57
tbodt has joined #pypy
14:57
yuyichao has quit [Ping timeout: 240 seconds]
14:57
mihaid has joined #pypy
15:07
mattip has joined #pypy
15:07
arigato has quit [Quit: Leaving]
15:08
realitix has joined #pypy
15:09
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
15:13
egregius313 has quit [Remote host closed the connection]
15:14
tbodt has joined #pypy
15:14
egregius313 has joined #pypy
15:16
tbodt has quit [Client Quit]
15:18
egregius313 has quit [Remote host closed the connection]
15:18
egregius313 has joined #pypy
15:22
egregius313 has quit [Remote host closed the connection]
15:23
egregius313 has joined #pypy
15:26
John has joined #pypy
15:26
<
John >
hi all, quick Q
15:27
<
John >
is it possible that manually compiling PyPy on OSX might give better performance over using the precompiled binaries?
15:27
<
John >
i have no evidence to suggest it does or doesn't, just a general question
15:27
egregius313 has quit [Ping timeout: 255 seconds]
15:28
<
kenaan >
antocuni faster-rstruct-2 43a0bd03dd95 /: close branch about to be merged
15:28
<
kenaan >
antocuni default d3895494491b /: merge the faster-rstruct-2 branch: this makes struct.pack and struct.pack_into much faster by using raw_store o...
15:28
Tiberium has quit [Remote host closed the connection]
15:28
<
kenaan >
antocuni default add017ddb1bc /pypy/doc/whatsnew-head.rst: document the merged branch
15:29
<
mattip >
antocuni: yay !
15:29
<
antocuni >
horray :)
15:29
jcea has joined #pypy
15:29
<
mattip >
I wonder how it will influence benchmark runs
15:30
<
antocuni >
I don't know, do we have benchmarks which rely on struct heavily?
15:30
tbodt has joined #pypy
15:31
<
mattip >
dunno also
15:32
tbodt has quit [Client Quit]
15:41
<
antocuni >
Krakatau seems to use struct.pack heavily in assembler.py
15:41
oberstet2 has joined #pypy
15:44
tbodt has joined #pypy
15:45
oberstet has quit [Ping timeout: 255 seconds]
15:49
* ronan
merging default into py3.5
15:50
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
15:50
<
mattip >
ronan: indeed, map(ord, x.__buffer__()) gives identical results on pypy 5.7 and cpython, but a memoryview object on HEAD
15:51
* mattip
trying to write a test with our array.c
15:51
<
ronan >
mattip: huh? x.__buffer__ doesn't exist on cpython
15:51
tbodt has joined #pypy
15:52
<
mattip >
correct, I meant it gives different results on pypy5.7 and HEAD
15:52
tbodt has quit [Client Quit]
15:53
tbodt has joined #pypy
15:53
<
ronan >
mattip: I tried to write tests with array.c but it doesn't seem to have the issue
15:54
<
ronan >
I don't understand why array.c and numpy.ndarray would behave differently though
15:54
<
mattip >
did you try -A tests?
15:56
<
mattip >
indeed, a simple -A tests with array.c does not show a difference between the two
15:57
tbodt has quit [Ping timeout: 240 seconds]
16:00
<
John >
oh, did the struct optimizations recently get added?
16:01
<
mattip >
John: yes, about 30 minutes ago :)
16:02
<
mattip >
John: you would have to tell us if you can improve the CFLAGS in the Makefile used to compile PyPy on OS X
16:03
<
John >
aww sweet - i think i've been waiting for this for two years :D
16:03
<
John >
mattip - ah, ok, so probably not :P
16:03
<
mattip >
ronan: here is the test I am using, note the Exception on line 48
16:04
<
John >
I thought you might say 'of course - with knowledge of your specific type of laptop/etc, things can be optimised'
16:04
<
John >
but if that's not the case, never mind, i've drunk too much of the gentoo coolaid
16:06
<
John >
How do i test the new struct_pypy? :)
16:06
<
John >
i mean, can i use the binary on the website right now?
16:07
<
cfbolz >
John: nope, you need to wait for the next nightly build (ie tomorrow)
16:07
<
kenaan >
Dodan PyPy_profopt_enabled 8fe61093e999 /: PyPy 5.8.0 functional profopt. Precleanup
16:07
<
kenaan >
Dodan PyPy_profopt_enabled f1aa2fc615ef /: PyPy-5.8.0-Profopt-enabled-postclean
16:07
<
kenaan >
Dodan PyPy_profopt_enabled ee2fb3e0d9b6 /pypy/goal/regrtest/: PyPy-5.8.0-Profopt-enabled-postclean
16:07
<
kenaan >
Dodan PyPy_profopt_enabled b5c11d10c249 /: Profopt option now can come paired with --profoptpath to specify the training workload
16:07
<
John >
ahh, ok :) awesome!
16:08
<
John >
Really looking forward to it :)
16:08
<
mattip >
ronan: it seems that exception has been there at least since 5.7.0
16:35
<
kenaan >
rlamy py3.5 66108bb2353d /: hg merge default
16:36
Tiberium has joined #pypy
16:37
Tiberium has quit [Remote host closed the connection]
16:40
inad922 has joined #pypy
16:46
realitix has quit [Ping timeout: 240 seconds]
16:47
<
tos9 >
mattip: out of curiosity, was there something specific that made you turn that big endian thing into a blocker
16:48
<
tos9 >
mattip: does numpy or something use it and now you noticed it was broken
16:50
yuyichao has joined #pypy
16:54
cstratak_ has quit [Ping timeout: 240 seconds]
16:55
<
ronan >
hrm, how does one install a debug version of numpy, again?
16:57
pilne has joined #pypy
17:06
<
kenaan >
stefanor py3.5 ad5a0fedc353 /lib_pypy/pyrepl/: Everything in lib_pypy should at least be compatible with py3k syntax
17:09
<
mattip >
ronan: rm -rf build; CFLAGS='-g -O0' pypy setup.py install
17:10
<
mattip >
tos9: there is a failing test, and it seems important, no?
17:10
<
mattip >
maybe now it will get more attention
17:16
<
ronan >
mattip: that doesn't seem to work. I can't see the source in gdb
17:17
<
mattip >
strange, maybe rm site-packages/numpy* ?
17:18
<
mattip >
ronan: you should be able to see the compilation flags fly by on the gcc compile command lines
17:18
<
ronan >
mattip: yes, they look right: C compiler: cc -pthread -DNDEBUG -O2 -O0 -g -fPIC
17:19
<
mattip >
maybe it is not being installed to the right place? strange ...
17:21
<
mattip >
when I type >>>>np.core.multiarray, I get the path to the multiarray.pypy-41.so, then I can see when it was built
17:22
* mattip
trying a build with DEBUG_WRAPPER to print calls to functions
17:23
oberstet2 has quit [Ping timeout: 255 seconds]
17:27
<
ronan >
mattip: solved it: I was running from numpy's root dir, which breaks the world in weird ways
17:29
marky1991_2 has joined #pypy
17:29
marky1991_2 has quit [Remote host closed the connection]
17:30
marky1991_2 has joined #pypy
17:31
marky1991 has quit [Ping timeout: 246 seconds]
17:34
<
ronan >
mattip: I see what's going on: numpy's array_getbuffer checks the flags and fills certain fields (most critically ndim) only if they have been requested
17:36
<
ronan >
however, our code always trusts view->ndim, and we end up with a 0-dim view
17:38
<
ronan >
I think that the bug was already there, but hidden because we (incorrectly, AFAICT) gave priority to getreadbuffer
17:39
<
mattip >
our PyBuffer_FillInfo also checks this flag
17:42
inad922 has quit [Ping timeout: 260 seconds]
17:44
<
ronan >
I have to go, I'll try to fix it tonight
17:45
<
mattip >
cool, thanks for digging into this
17:45
marky1991 has joined #pypy
17:47
marky1991_2 has quit [Ping timeout: 240 seconds]
17:49
oberstet2 has joined #pypy
18:16
lvella has joined #pypy
18:17
<
lvella >
hi, is is possible to return a callable python object as a c function pointer using CFFI?
18:18
<
lvella >
how is the object lifetime managed?
18:39
<
lvella >
just did, doesn't seem pertinent, because "The @ffi.def_extern() decorator should be applied to global functions"
18:39
<
lvella >
I want to use a callable object, not a global function
18:39
<
tos9 >
mattip: I mean, I agree :), I think the old behavior was potentially masking lots of bugs -- was just curious. Cool
18:41
<
mattip >
lvella: sorry, can't really help without seeing the code, and even then probably will end up not being helpful
18:41
<
LarstiQ >
lvella: callable global object?
18:42
<
LarstiQ >
lvella: or are you gettting callables you can't predict at compile time?
19:07
<
lvella >
LarstiQ: imagine I created an arbitrary callable local object, maybe with functools.partial
19:08
<
lvella >
is it possible to take a C function pointer from it, that can be directly called from C?
19:14
tbodt has joined #pypy
19:21
jamescampbell has joined #pypy
19:31
egregius313 has joined #pypy
19:35
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
19:35
tbodt has joined #pypy
19:37
jamescampbell has quit [Remote host closed the connection]
19:40
egregius313 has quit [Remote host closed the connection]
19:41
yuyichao has quit [Ping timeout: 255 seconds]
19:41
egregius313 has joined #pypy
20:04
nphg has joined #pypy
20:22
egregius313 has quit [Remote host closed the connection]
20:34
ramonvg has joined #pypy
20:37
jamescampbell has joined #pypy
20:42
jamescampbell has quit [Ping timeout: 268 seconds]
20:46
kolko has joined #pypy
21:00
lvella has quit [Quit: Page closed]
21:02
ramonvg has quit [Ping timeout: 245 seconds]
21:04
kipras is now known as kipras`away
21:20
tav has quit [Quit: tav]
21:21
arigato has joined #pypy
21:27
egregius313 has joined #pypy
21:39
arigato has quit [Quit: Leaving]
21:40
adeohluwa has joined #pypy
22:06
egregius313 has quit [Remote host closed the connection]
22:12
egregius313 has joined #pypy
22:23
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
22:26
tbodt has joined #pypy
22:31
egregius313 has quit [Remote host closed the connection]
22:32
amaury has quit [Ping timeout: 246 seconds]
22:49
jamescampbell has joined #pypy
22:50
yuyichao has joined #pypy
22:53
jamescampbell has quit [Quit: Leaving...]
23:11
ESphynx has left #pypy [#pypy]
23:15
antocuni has quit [Ping timeout: 268 seconds]
23:17
egregius313 has joined #pypy
23:27
egregius313 has quit [Remote host closed the connection]
23:28
egregius313 has joined #pypy
23:29
egregius313 has quit [Remote host closed the connection]
23:29
egregius313 has joined #pypy
23:31
egregius313 has quit [Remote host closed the connection]
23:32
egregius313 has joined #pypy
23:35
yuyichao has quit [Ping timeout: 260 seconds]
23:36
egregius313 has quit [Ping timeout: 246 seconds]
23:50
nphg1 has joined #pypy
23:52
nphg has quit [Ping timeout: 240 seconds]
23:57
marr has quit [Ping timeout: 272 seconds]