brechtm has quit [Remote host closed the connection]
brechtm has joined #pypy
asmeurer__ has joined #pypy
asmeurer__ has quit [Client Quit]
jamesaxl has joined #pypy
jamesaxl has quit [Client Quit]
antocuni has joined #pypy
vkirilichev has joined #pypy
oberstet3 has quit [Ping timeout: 260 seconds]
<kenaan>
tobweber stmgc[c8-long-transactions] df52953567cb /c8/stm/: Define large nursery mark in a more straight forward way and make single thread mode transac...
brechtm has quit [Remote host closed the connection]
black_ant has quit [Ping timeout: 240 seconds]
black_ant has joined #pypy
antocuni has quit [Ping timeout: 260 seconds]
<kenaan>
arigo default d75b6c67c8e3 /lib_pypy/_ctypes/: Semi-blind fix for issue #2533. Did not manage to extract a simple test (but admittedly, gave up after 15 minutes)
brechtm has joined #pypy
brechtm has quit [Remote host closed the connection]
brechtm has joined #pypy
demonimin has quit [Ping timeout: 252 seconds]
brechtm has quit [Ping timeout: 260 seconds]
<kenaan>
arigo default da1964eea7ad /pypy/module/test_lib_pypy/ctypes_tests/test_functions.py: Managed to write a test for d75b6c67c8e3.
demonimin has joined #pypy
arigato has quit [Quit: Leaving]
demonimin_ has joined #pypy
demonimin_ has joined #pypy
demonimin has quit [Ping timeout: 260 seconds]
brechtm has joined #pypy
demonimin_ has quit [Ping timeout: 240 seconds]
ramon_ has joined #pypy
ramon_ is now known as ramonvg
lritter has joined #pypy
black_ant has quit [Ping timeout: 268 seconds]
jkbbwr has quit [Ping timeout: 260 seconds]
necaris has joined #pypy
necaris is now known as necaris[away]
necaris[away] is now known as necaris
necaris is now known as necaris[away]
jamesaxl has joined #pypy
adamholmberg has joined #pypy
necaris[away] has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jamesaxl has quit [Read error: Connection reset by peer]
jamesaxl has joined #pypy
jamesaxl has quit [Client Quit]
jamadden has joined #pypy
jamesaxl has joined #pypy
jamesaxl has quit [Read error: Connection reset by peer]
jamesaxl has joined #pypy
adamholmberg has quit [Remote host closed the connection]
adamholmberg has joined #pypy
brechtm_ has joined #pypy
adamholmberg has quit [Ping timeout: 240 seconds]
brechtm has quit [Ping timeout: 240 seconds]
sven__ has quit []
jamesaxl has quit [Read error: Connection reset by peer]
<johncc3_>
I can't locate the problem. The structure seems right to me.
<fijal>
struct snd_pcm_t *
<fijal>
where is that structure defined?
<danchr>
is there a bug concerning PyPy's somewhat nonstandard packing layout?
<danchr>
(e.g. $prefix/site-packages)
<johncc3_>
fijal, It's defined in the ALSA .h's
<danchr>
s/packing/packaging)
<fijal>
johncc3_: right, so you need to say at least "typedef struct snd_pcm_t { ...; }
<arigato>
johncc3_: ffi.cdef() is not for C *code*
<fijal>
"
<fijal>
and the output stream need to be declared as a declaration only too :-)
<danchr>
arigato: your fix made ssl compile on my Mac, thanks!
<arigato>
danchr: :-)
* danchr
is trying to packaging pypy3 for MacPorts
FASDFSF has joined #pypy
<danchr>
has anyone tried to change the packaging layout to more closely resemble CPython? would that be an acceptable change?
<arigato>
johncc3_: .cdef() is for defining which types and functions need to be accessible from Python, but you can't *implement* anything in a cdef
<johncc3_>
arigato, Thanks. Where _do_ I put the C-code then?
<fijal>
in set_source()
<LarstiQ>
danchr: in what way?
<johncc3_>
fijal, I tried that, but got a load of other errors. Will try again.
<fijal>
johncc3_: the whole thing is a bit misguided, cdef needs to have declarations and set_source C code
<arigato>
what you put in set_source() must be valid C code, with little automatic magic added
<arigato>
it's not even parsed by cffi or anything
<arigato>
just sent to the C compiler
<johncc3_>
arigato, Great - moved C code. I now get warning: function declaration isn’t a prototype [-Wstrict-prototypes]
<johncc3_>
Do I have to add separate prototypes elsewhere?
<danchr>
LarstiQ: so that it uses $prefix/lib, for example
<arigato>
johncc3_: that's gcc complaining that you wrote "struct snd_pcm_t *open_output_stream()"
<LarstiQ>
danchr: is the value of $prefix the same?
<danchr>
$prefix is whereever you installed PyPy
<arigato>
it gives you a warning because it's really supposed to be "struct snd_pcm_t *open_output_stream(void)"
<arigato>
(that's all standard C, even if C is sometimes obscure)
<danchr>
LarstiQ: in my case, /opt/pypy2 for custom installation and /opt/local for MacPorts
<johncc3_>
arigato, typedef int... snd_pcm_uframes_t; I understand the 'int...' is correct literally like this?
<LarstiQ>
danchr: is that an OSX thing? Never seen $python-dir/lib
<arigato>
yes, that's a cffi-specific C extension for the cdef()
<johncc3_>
arigato, I get "error: expected identifier or ‘(’ before ‘...’ token"
<arigato>
can't help, need the complete paste
<arigato>
or did you move "typedef int... snd_pcm_uframes_t;" to set_source()?
<arigato>
don't do that, that's not standard C
vkirilic_ has joined #pypy
<arigato>
as I said, that's a cffi-only C extension _for the cdef_
<danchr>
LarstiQ: a normal CPython 2.7 build, for example, places it's Python sources within $prefix/lib/python2.7 — where $prefix defaults to /usr/local, and is /usr for in a typical Linux distribution
<LarstiQ>
danchr: ah, so I misunderstood your earlier value of $prefix
<LarstiQ>
danchr: yes
<danchr>
(CPython is slightly odd on OS X, as it builds it's library as a framework, which is a Mac-specific thing, but that's a separate thing)
vkirilichev has quit [Ping timeout: 240 seconds]
<danchr>
gah, translating pypy is really slow :-)
<danchr>
I don't suppose someone succeeded in parallelising it? ;-)
johncc3_ has quit [Ping timeout: 240 seconds]
<nimaje>
afaik the main part of translating is python code and after generating c make is called with -j
<danchr>
yeah, most of it is black magic
johncc3_ has joined #pypy
black_ant has joined #pypy
jamesaxl has quit [Ping timeout: 240 seconds]
jamesaxl has joined #pypy
black_ant has quit [Ping timeout: 240 seconds]
necaris is now known as necaris[away]
jamesaxl has quit [Read error: Connection reset by peer]
necaris[away] has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jamesaxl has joined #pypy
black_ant has joined #pypy
brechtm has joined #pypy
jamesaxl has quit [Read error: Connection reset by peer]
<haypo>
ronan: i also wrote a small tool to plot values. example of 1 sympy_sum run on CPython, 1 warmup and 50 values: http://www.haypocalc.com/tmp/sympy_sum.png (warmup not rendered here)
<haypo>
it _looks_ like a cycle ;)
jeroud has joined #pypy
<gsnedders>
haypo: I wonder if it's something with the ridiculously polymorphic call-sites in html5lib
<gsnedders>
esp. given they'll grow in number over test iterations because of stupid design decisions (which, okay, might not have been quite so stupid originally)
<ronan>
haypo: another weird thing is that on CPython, there are benchmarks with a lot of variation between runs, but not within runs
<haypo>
ronan: i noticed that on microbenchmarks like call_simple, because of different number of hash collisions (because of the randomized hash function)
black_ant has quit [Ping timeout: 240 seconds]
<haypo>
ronan: i should try to run benchmarks with much more values per run and take a look at values :)
<haypo>
ronan: at least, the spikes are deterministic :)
<mattip>
haypo: nice
<ronan>
haypo: nice graph! No clue what it actually means...
adamholmberg has quit [Remote host closed the connection]
brechtm has quit [Read error: Connection reset by peer]
brechtm has joined #pypy
<mattip>
IMO it means that a benchmark runner should not invoke heruistics to determine if one version of the interpreter is better than another
<mattip>
rather the benchmark run parameters should be very repeatable, and the raw data for each cycle of each benchmark run should be available for analysis
<mattip>
s/heruistics to determine/heruistics to run benchmarks when determining/
Tiberium has joined #pypy
<haypo>
ronan: at least, on sympy_sum it's more stable if you disable the GC :-p
<jamadden>
http://pypy.org/performance.html says that "Classes that inherit from both new- and old-style classes are extremely slow." Does anyone know if that's still the case on PyPy2 5.7.1? I'm running a large application benchmark that seems to support that it is still true (when I flip some key classes from mixed to new-style it gets faster) but the benchmark is synthetic, and there's enough noise I'm not completely sure
<haypo>
mattip: FYI all i said last our was about CPython :) on CPython, the number of warump is fixed to 1
asmeurer has joined #pypy
<haypo>
minor enhancement: I removed call_method, call_simple, call_* and pybench from the performance project and moved them to my https://github.com/haypo/pymicrobench project ;-)
<cfbolz>
jamadden: new style is definitely much better optimized
<cfbolz>
old style classes aren't totally terrible anymore, but still not as good
<jamadden>
cfbolz: Thanks!
[Arfrever] has joined #pypy
<cfbolz>
jamadden: if you have a concrete real life problem with old style classes we can think about improving
<jamadden>
cfbolz: I'd much rather get the old style classes out of the application :) I think that's do-able since it runs successfully on Python 3 with the same classes
<cfbolz>
Then it should Just Work (tm)
asmeurer_ has joined #pypy
ramonvg has joined #pypy
jcea has quit [Quit: jcea]
asmeurer_ has quit [Quit: asmeurer_]
bgola_ is now known as bgola
adamholmberg has quit [Remote host closed the connection]
brechtm has quit [Read error: Connection reset by peer]
brechtm_ has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
brechtm has joined #pypy
* mattip
honing in on issue 2530
adamholmberg has joined #pypy
<mattip>
issue 2530 seems to involve converting list -> ndarray(..., dtype=object) -> ndarray(..., dtype=int64)
brechtm_ has quit [Ping timeout: 252 seconds]
<mattip>
when using multiple threads
<jamadden>
cfbolz: It does appear to Just Work. And when I get stable numbers, the performance difference is impressive: 2.5X on some benchmarks. (https://github.com/zopefoundation/ZODB/pull/160) Thanks again.
ionelmc has quit []
ionelmc has joined #pypy
johncc3_ has quit [Ping timeout: 260 seconds]
jkbbwr has left #pypy [#pypy]
shodan45_ has quit [Remote host closed the connection]
shodan45 has joined #pypy
johncc3_ has joined #pypy
jcea has joined #pypy
kipras`away is now known as kipras
ramonvg has quit [Ping timeout: 260 seconds]
marr has joined #pypy
jcea has quit [Read error: Connection reset by peer]
jcea1 has joined #pypy
jcea1 is now known as jcea
bplotnick has joined #pypy
brechtm_ has joined #pypy
brechtm has quit [Ping timeout: 240 seconds]
Rhy0lite has quit [Quit: Leaving]
demonimin has joined #pypy
demonimin has joined #pypy
<mattip>
perhaps somehow an ob_pypy_link is not being updated when a W_Root object is moved?
<mattip>
the test creates long lists and then calls PyList_GetItem on them, which returns a borrowed ref
<mattip>
mabe they are not tracked well when moved by the GC ?
vkirilichev has joined #pypy
* mattip
still trying to minimize the test
necaris is now known as necaris[away]
necaris[away] has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
necaris has joined #pypy
<mattip>
somehow connected to cython translation, pure python works AFAICT
* mattip
zzz
mattip has left #pypy ["bye"]
johncc3_ has quit [Ping timeout: 240 seconds]
h264 has joined #pypy
Tiberium has quit [Remote host closed the connection]
<h264>
hi\
<h264>
libpypy-c.so is missing
<h264>
what can i do?
marky1991 has quit [Ping timeout: 260 seconds]
johncc3_ has joined #pypy
tbodt has joined #pypy
adamholmberg has quit []
brechtm_ has quit [Remote host closed the connection]
adamholmberg has joined #pypy
brechtm has joined #pypy
jcea has quit [Remote host closed the connection]
jcea has joined #pypy
vkirilichev has quit [Remote host closed the connection]
a-tal has quit [Remote host closed the connection]
adamholmberg has quit [Remote host closed the connection]
adamholmberg has joined #pypy
FASDFSF has quit [Read error: Connection reset by peer]
bplotnick has quit [Remote host closed the connection]
bplotnick has joined #pypy
adamholmberg has quit [Ping timeout: 260 seconds]
nimaje1 has joined #pypy
nimaje1 is now known as nimaje
nimaje is now known as Guest34685
bplotnick has quit [Remote host closed the connection]