tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has joined #pypy
tbodt has quit [Client Quit]
<njs>
liqian: how did you install this library? a library with "cpython" in the name should not be referencing "PyPyExc_TypeError".
<njs>
liqian: reinstalling cython won't help because cython isn't the library with the problem; you need to reinstall pycrypto (I'm guessing)
<njs>
(though also fyi pycrypto is not recommended for general use anymore)
<liqian>
reinstall pycrypto doesnt work
<liqian>
i tried
<liqian>
reinstall still have the error
<liqian>
i guess may be someting enviroment may cause this??
<njs>
liqian: you might have more luck in #python, since it doesn't sound like pypy is involved except accidentally, and there are a lot more people in there...
<njs>
but yes, it sounds like something about the way you're installing, or the environment that you're trying to install into, is broken
<liqian>
thanks~
liqian has quit [Quit: Leaving]
tbodt has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has joined #pypy
tilgovi has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has joined #pypy
_whitelogger has joined #pypy
adw1n has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has joined #pypy
tbodt has quit [Client Quit]
pilne has quit [Quit: Quitting!]
tbodt has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has joined #pypy
tbodt has quit [Client Quit]
jamadden has quit [Quit: Leaving.]
tbodt has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
Kronuz has quit [Ping timeout: 272 seconds]
tbodt has joined #pypy
tbodt has quit [Client Quit]
adw1n has quit [Remote host closed the connection]
tbodt has joined #pypy
Kronuz has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
vkirilichev has joined #pypy
<kenaan>
arigo cffi/cffi[char16_char32_t] 68b58f5621f8 /: ready to merge
<kenaan>
arigo cffi/cffi 719b689e9e4b /: hg merge char16_char32_t Issue #315: add 'char16_t' and 'char32_t', which are explicitly-sized vresions of 'wcha...
<kenaan>
arigo cffi/cffi c60281bf502f /doc/source/: Document char16_t and char32_t
arigato has joined #pypy
<kenaan>
arigo cffi/cffi 19722530a814 /testing/cffi1/test_recompiler.py: Don't pass '-Werror' to MSVC
<realitix>
hello arigato! I have been selected to PyCon AU (australia). I have two questions for you. 1 - With ABI mode, when you use set_source, it's only for out-of-line module generation ? You don't need set_source with in-line mode ? 2 - what is the purpose of libffi ? Is it used only to create C type ? When you do ffi.new(...)
jacob22_ has joined #pypy
cstratak has joined #pypy
<kenaan>
arigo cffi/cffi c0c2020872d3 /: Fix: the type char32_t is defined by the standard as always unsigned
<kenaan>
arigo cffi/cffi 46cd2faf0897 /c/test_c.py: An extra test to distinguish wchar_t from char32_t, at least on linux
[Arfrever] has joined #pypy
<kenaan>
arigo cffi/cffi 6c855193e59c /doc/source/ref.rst: Document the signedness of charN_t
<GrimFand>
Hi there, I found out that vmprof can trace native C functions on Linux. I tried to find out how this is achieved, and it looks like there is a dependency on an external library called "libunwind". I am currently playing around with the frame evaluation function introduced to Python 3.6 and if I can somehow print out more information when my callback function is called when entering a pure C...
<GrimFand>
...module function.
<antocuni>
GrimFand: what is the question? :)
<arigato>
realitix: 1 - yes, you're correct. 2 - maybe your question was truncated?
<GrimFand>
my goal is to see which C functions are called by all my Python statements.
<antocuni>
GrimFand: before they are called?
<GrimFand>
so if I run print("Hello World") in my interpreter console, then it should print out all the C functions called.
<GrimFand>
when I print out PyFrameObject->f_code->co_name for this case it just prints out "<module>"
<antocuni>
there is no way to do that; vmprof (and libunwind) works differently: you can stop your program at any time and see what is the current stacktrace (roughly the equivalent of running it inside gdb, press ctrl-c and type "where")
<arigato>
GrimFand: *all* the C function that are ever called? or are we talking about a sampling profiler here too?
<antocuni>
but you cannot ask "tell me all the C functions which will be called from now on"
<antocuni>
GrimFand: also, f_code->co_name == "<module>" means that it's python code at the top level of a module, not C code
<GrimFand>
If I look at sys.settrace, which used to be the main choice for printing out the called functions, there is an event c_call described in the documentation. So I expect that Python core developers intended to support tracing of C extension functions.
<GrimFand>
antocuni: Ok
<antocuni>
GrimFand: c_call tells you that python is calling a C function; then, the C function can call whatever else it wants but you won't be noticed at all
<GrimFand>
arigato: I just want to make my python interpreter verbose: I want to know exactly which Python C API functions get called when I call a specific Python function
<antocuni>
GrimFand: you might be interested in ltrace
<LarstiQ>
GrimFand: note that it might not call Python C API functions at all
<GrimFand>
antocuni: I would like to get this implemented across platforms
<LarstiQ>
but maybe this was not what you meant
<antocuni>
GrimFand: what you want to do is simply not possible
<GrimFand>
I was thinking about somehow getting the address of the currently called function and then resolve the symbol name with dladdr on Linux, and Win32 API functions on Windows
mattip has joined #pypy
<antocuni>
mattip: hi
<mattip>
hi
<antocuni>
what is the roadmap to pypy 5.8?
<arigato>
hi
<GrimFand>
but first of all I need to somehow register a callback function that gets called every time not only a pure Python function frame is entered, but also every C function framem which is probably not possible by means of PEP 523 (frame evaluation API)
<mattip>
pypy2 seems ready, except for a failing test on cffi 1.11 on OS X that I wanted to point out to arigato :)
<mattip>
arigato: hi
<arigato>
mattip: unless you want to wait more, I'd say that cffi should contain 1.10 in pypy 5.8
<antocuni>
mattip: I just discovered that vmprof hangs on pypy 5.8 but not on 5.7.1 :(
<mattip>
arigato: ok, I will try to backout those changes, but then the comparefile test will fail of course
<arigato>
yes
<arigato>
I think we can ignore this one
<mattip>
right
<gabrielm>
hi, I studied a little bit the StringBuilder and I saw that in the test files, after a build you can do append but if I try this in a normal python script, it fails. This behavior will remain like this from now on, in python scripts?
<antocuni>
specifically, it hangs when calling _vmprof.disable(), which means that we need to fix something in pypy :(
<mattip>
ok, so I guess that is a blocker for the release. Nice catch
<antocuni>
yeah, I fear so
<mattip>
better before than after
<antocuni>
lets see what plan_rich thinks when he gets online
<mattip>
ok
<mattip>
I am a bit sad that again we will release only linux64 PyPy3.5, the others are millimeters away from working
<antocuni>
yeah :(
<arigato>
gabrielm: I don't know why there are such tests, because a stringbuilder that got build() called on it doesn't work any more, after translation
<mattip>
pareto principle strikes again
<gabrielm>
In Java, a StringBuilder will still work after you call build on it
<mattip>
OTOH, down to 8 failing numpy tests, here is a summary (the last lines)
<mattip>
ok, I'll try the tests on pandas HEAD, they merged a fix for us
<mattip>
LarstiQ: assuming you are interested in pandas, right?
<LarstiQ>
mattip: ah, I was thinking of this numpy progress
<LarstiQ>
mattip: I've been treating that url as an "how is pypy scientific stack progressing"
jacob22_ has quit [Quit: Konversation terminated!]
<mattip>
LarstiQ: it's more like me just documenting rolling the rock up the hill
<mattip>
user cases of real code that doesn't work or doesn't work fast would be much more helpful
<kenaan>
arigo default 2dbc1ac2dab9 /pypy/module/__pypy__/: __pypy__.StringBuilder: after build(), we can continue to append more strings to the same builder. This is support...
<arigato>
gabrielm: fixed in 2dbc1ac2dab9 by removing the check in the __pypy__ module
<arigato>
thanks for pointing it out
<njs>
mattip: are there nightly builds of pypy3.5 on anything besides linux64? I should add them to my CI at some point...
<LarstiQ>
mattip: I might start actually working with that in the coming months, in which case I certainly will try help you roll the rock
<mattip>
njs: ARM seems to be building
<njs>
oh uh. that's neat but I don't have any ARM in my CI :-)
<mattip>
:)
<arigato>
mattip: win32 py3.5 build seems stuck to an older revision
<arigato>
mattip: ah no, this new failure is very recent and was fixed,
<arigato>
but there used to be a different failures the days before
<arigato>
right, the gen_makefile() failure should have been fixed by merging default
<mattip>
ok, sorry for the noise, it's just the slow windows buildbot lagging behind the others
<mattip>
on to making cffi 1.10 for the release
<mattip>
backing out each commit seems too painful
<mattip>
I will just revert it all to a <known good version>
<arigato>
mattip: I think you should just revert the merge of cffi-complex
<mattip>
ok, trying
nimaje1 has joined #pypy
nimaje1 is now known as nimaje
nimaje has quit [Killed (wilhelm.freenode.net (Nickname regained by services))]
<arigato>
there was no other changes to module/_cffi_backend recently
<mattip>
"abort: cannot backout a merge changeset"
<kenaan>
arigo cffi/cffi f4ff36763712 /doc/source/whatsnew.rst: Mention that 0.10.1 is only inside PyPy 5.8.0
<kenaan>
arigo release-pypy2.7-5.x 0ec86a2b43ab /pypy/module/test_lib_pypy/cffi_tests/cffi0/test_parsing.py: I think this test needs to change too (from 'hg diff -r 486d919c0b87 -r 93fa52b7eed3 testing/' inside c...
<kenaan>
arigo release-pypy3.5-5.x 1948ff94a48d /pypy/module/test_lib_pypy/cffi_tests/cffi0/test_parsing.py: I think this test needs to change too (from 'hg diff -r 486d919c0b87 -r 93fa52b7eed3 testing/' inside c...
<kenaan>
arigo release-pypy2.7-5.x 1bfb11c86cfd /: Give the version 1.10.1 to cffi
<kenaan>
arigo release-pypy3.5-5.x c29e21f6dd75 /: Give the version 1.10.1 to cffi (grafted from 1bfb11c86cfd81a86d54823f85aad0ca5d89fd98)
arigato has quit [Quit: Leaving]
arigato has joined #pypy
marky1991 has joined #pypy
jamesaxl has joined #pypy
tbodt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
jamesaxl has quit [Read error: Connection reset by peer]
jamesaxl has joined #pypy
yuyichao has joined #pypy
rmesta has joined #pypy
rmesta has left #pypy [#pypy]
marky1991 has quit [Ping timeout: 255 seconds]
jamesaxl has quit [Read error: Connection reset by peer]
<kenaan>
arigo cffi/cffi fd8ab01804fb /c/test_c.py: an extra test