Mailman 3 January 2013 - cython-devel

Re: [Cython] Bringing Cython and PyPy closer together
by Stefan Behnel March 29, 2015

March 29, 2015

[copied here from PyPy mailing list] Stefan Behnel, 15.02.2012 12:32: > The current state of the discussion seems to be that PyPy provides ways to > talk to C code, but nothing as complete as CPython's C-API in the sense > that it allows efficient two-way communication between C code and Python > objects. Thus, we need to either improve this or look for alternatives. > > In order to get us more focussed on what can be done and what the > implications are, so that we may eventually be able to decide what should > be done, I started a Wiki page for a PyPy backend CEP (Cython Enhancement > Proposal). > > http://wiki.cython.org/enhancements/pypy The discussion so far makes me rather certain that the most promising short-term solution is to make Cython generate C code that PyPy's cpyext can handle. This should get us a rather broad set of running code somewhat quickly, while requiring the least design-from-scratch type of work in a direction that does not yet allow us to see if it will really make existing code work or not. On top of the basic cpyext interface, it should then be easy to implement obvious optimisations like native C level calls to Cython wrapped functions from PyPy (and potentially also the other direction) and otherwise avoid boxing/unboxing where unnecessary, e.g. for builtins. After all, it all boils down to native code at some point and I'm sure there are various ways to exploit that. Also, going this route will help both projects to get to know each other better. I think that's a required basis if we really aim for designing a more high-level interface at some point. The first steps I see are: - get Cython's test suite to run on PyPy - analyse the failing tests and decide how to fix them - adapt the Cython generated C code accordingly, special casing for PyPy where required Here is a "getting started" guide that tells you how testing works in Cython: http://wiki.cython.org/HackerGuide Once we have the test suite runnable, we can set up a PyPy instance on our CI server to get feed-back on any advances. https://sage.math.washington.edu:8091/hudson/ So, any volunteers or otherwise interested parties to help in getting this to work? Anyone in for financial support? Stefan

2 9

Re: [Cython] [cython-users] Recommendations for efficient typed arrays in Cython?
by Sturla Molden Feb. 4, 2013

Feb. 4, 2013

On 29.01.2013 22:15, Sal wrote: > I don't have any 'typical' Python code yet, everything is 'cdef' type > stuff, but reference counting is still performed by Cython on 'cdef' > types, so sticking them into a C++ array probably ignores that fact. I think the main difficulty with Cython (for beginners at least) is knowing 'when we are in C land' and 'when we are in Python land'. I think this is the fundamental problem you are facing here. I therefore think it is is unfortunate that we write cdef object a cdef list b cdef foobar c etc to define Python variables. 'cdef' seems to indicate that it is a C declaration, yet here it is not. Neither does this cdef syntax allow us to declare Python int and float statically. Perhaps it would be easier if Python variables were declared with 'pydef'? Or perhaps just 'def'? Then it might be easier to see what is Python and what is C. Sturla

3 6

embedsignature issue (with both Sphinx and Epydoc)
by g b Jan. 31, 2013

Jan. 31, 2013

Hello, I tried the "embedsignature=True" compilation directive on both Sphinx and Epydoc and it resulted in : I had to pass a more python compliant signature in the very first line of the docstring for each method of my classes. See this thread : http://stackoverflow.com/questions/14616882/how-to-make-epydoc-show-paramet…. Exported signature is of the form : PyLabNode.SetNetwork(self, PyLabNetwork net) This has 2 drawback : The dotted notation for the class prefix and the typed parameter. It seems to me that its feature doesn't do what it was designed for. Or I'm I missing something... Thanks.

1 0

Re: [Cython] [cython-users] Nightly builds in "Kivy Daily" PPA
by Stefan Behnel Jan. 29, 2013

Jan. 29, 2013

Thomas Pietrowski, 29.01.2013 20:54: > As I'm packaging for the Kivy team and we needed updated packages for > Cython I made nightly builds using recipes at launchpad. > Which means everytime you, the cython devs, commit changes at github a > package gets build in our PPA. > > https://launchpad.net/~kivy-team/+archive/kivy-daily > > We had the discussion before and I would be happy I you agree to add a PPA > to your team at launchpad and get these packages built there. > I would of course help you getting it work, I just need a PPA in the team > to target the recipes to get built there. I have no experience with this, but I just created a new PPA called "master-ppa". https://launchpad.net/~cython-dev/+archive/master-ppa If there's anything else I have to do, just let me know. Stefan

1 0

Cython 0.18 released
by Stefan Behnel Jan. 29, 2013

Jan. 29, 2013

Hi all, Cython 0.18 is out of the door. This is a feature release that adds some long requested language features (const!), better Python compatibility and optimisations. You can get it from here: http://cython.org/release/Cython-0.18.tar.gz http://cython.org/release/Cython-0.18.zip Release notes: https://github.com/cython/cython/blob/0.18/CHANGES.rst Documentation: http://docs.cython.org/ The major new features are: * A new directive ``overflowcheck`` which raises an OverflowException when arithmetic with C ints overflows. This has a modest performance penalty, but is much faster than using Python ints. * C ``const`` declarations are supported in the language. * Calls to nested Python functions are resolved at compile time. * Type inference works across nested functions. * ``py_bytes_string.decode(...)`` is optimised. * Named Unicode escapes ("\N{...}") are supported. * Python functions/classes provide the special attribute "__qualname__" as defined by PEP 3155. Have fun, Stefan

1 0

0.18 rc1 released
by Stefan Behnel Jan. 28, 2013

Jan. 28, 2013

Hi, here is a release candidate for 0.18. http://www.cython.org/release/Cython-0.18rc1.tar.gz http://www.cython.org/release/Cython-0.18rc1.zip The changes since the beta version were mostly fixes for the bugs reported so far. Please give it another try, also to see if we managed to fix what you found. Thanks! Stefan

2 1

http://docs.cython.org/dev/
by Stefan Behnel Jan. 27, 2013

Jan. 27, 2013

Hi, I set up a redirect from http://docs.cython.org/dev/ to the latest docs page built by Jenkins. Certainly easier to remember than the lengthy real URL. If anyone has a good idea where to show that link in the original (released) documentation, or how to make Sphinx insert it for us, I'd be happy to hear about it. A pull request would be well appreciated. :) Stefan

1 0

`_save` was not declared when cathing C++ exceptions in nogil cdef function
by Matěj Laitl Jan. 26, 2013

Jan. 26, 2013

Hi again, another minor problem I've stumbled upon - when I call an "except +" function in a cdef ... nogil function, gcc fails to compile the .cpp file. Perhaps a missing PyThreadState *_save; declaration in case of nogil functions? Cython 0.18b1 3e37475ec5098eef2c421. Attached patch trivially extends cpp_exceptions_nogil test to show the problem. Cheers, Matěj

1 1

Re: [Cython] Feature Request: expand list/tuple in 'if (... in ...):' ...
by Christian Inci Jan. 24, 2013

Jan. 24, 2013

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 01/24/2013 03:44 PM, Dag Sverre Seljebotn wrote: > How do you know that INT_TUPLE isn't reassigned to something else? On 01/24/2013 03:49 PM, Stefan Behnel wrote: > This is not an optimisation as it might render the code incorrect. How can > Cython know that you'll never change INT_TUPLE at runtime? A good point. What's about something like final or const? > However, if you use a literal list or tuple in place, this already works, > i.e. the first example is equivalent to > > if j in (0x66, 0x67, 0x2e, > 0x36, 0x3e, 0x26, 0x64, > 0x65, 0xf2, 0xf3, 0xf0): > > You actually get a C switch statement for it. Thanks, this actually works. :-) > > If you prefer keeping the set of integer values at a separate place, this > should work: > > DEF INT_TUPLE = (0x66, 0x67, 0x2e, > 0x36, 0x3e, 0x26, 0x64, > 0x65, 0xf2, 0xf3, 0xf0) > > (mind the upper case "DEF"). But this doesn't work. That's why I'm using 'cdef tuple'. According to 'http://docs.cython.org/src/userguide/language_basics.html#compile-time-defi…', this will only work using int, long, float or str. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJRAVDuAAoJEDfL/pqB6n6NsCYP/1HgWbMRuSi4o5o0gIxeSpfl hkesrLlKe3sodTsWt3cSPGWBg5Y1pq25SUSCLCNssJruXTZ66FCTyMffc3TpEuei PUSKq+byFplYEmCMPEm1l2z2vBI4+c5JggcXwDLlmp9RR6ZWXft0tb6fg/4Tqwyf 76/l6jrJiI7DImeJwHKDfSs1bHWCIZYPIjQWkeC42kETFG1G+U/L7tfZ8+7oq0pN fXhRn+XK2ax3iMsswzgIBAuyywASvT2uqMI3/G+MnD7tCiRFZkzILEa3+igxqMYN H/C9N33XLPZ1UulX40c88LHe1i2fpLvu1IvWoYc42Tszkd8Wd+uASs78b1puFTk+ jeeBLQkGbMle+eCXWQLfnSdmvvn3yYdE1UgVxQ4RXONjol9PSI1r86xn2V97phmW C/t4r6174hv49wAG1MoMVwXOUJH3ylEBz2qJFq06nwKAU7891dd9O1tXvoVsTihe NYDk6zflZlPZi7FUMs66S//rYYrXY/ZkA3/NhD9Krbd/Z1cVFTN1J/ynoccMx8dY pjRz9A+oIzXw/6kvITmGx22iKZRSlhLYe32ZT1qiGk8N1nGAvlaVIqK9mpeQhECX 5CEPSj4+t6T7th7Nh/fMOqJomf0AfzCsuz4QYlYQ8Y+7zfwuOnC8AkkoIid6ac1e M4vXi93fgnp64RpxKkxS =Lgq1 -----END PGP SIGNATURE-----

2 1

Feature Request: expand list/tuple in 'if (... in ...):' ...
by Christian Inci Jan. 24, 2013

Jan. 24, 2013

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Subject: Feature Request: expand list/tuple in 'if (... in ...):' ... expand list/tuple in 'if (... in ...):' on compile-time if the list/tuple contains only static values. test1: list is expanded manually. test2: list is not expanded. The output of the testcase is: test1: timediff: 0.205533 test2: timediff: 17.567185 Testcase: from time import time cdef tuple INT_TUPLE = (0x66, 0x67, 0x2e, 0x36, 0x3e, 0x26, 0x64, 0x65, 0xf2, 0xf3, 0xf0) cpdef test(): cdef unsigned int i cdef unsigned short j # 'unsigned char j' won't work for some reason cdef double start_time start_time = time() for i in range(0x50000): for j in range(256): if (j == 0x66 or j == 0x67 or j == 0x2e or j == 0x36 or j == 0x3e or j == 0x26 or j == 0x64 or j == 0x65 or j == 0xf2 or j == 0xf3 or j == 0xf0): pass print("test1: timediff: {0:f}".format(time()-start_time)) start_time = time() for i in range(0x50000): for j in range(256): if (j in INT_TUPLE): pass print("test2: timediff: {0:f}".format(time()-start_time)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJRAUGQAAoJEDfL/pqB6n6NG90P/jWKAw1vjEO1cyem2teyYIce cGWbewGJ14VJycztcnZG6XSNzIC0lLMbeu5edUwmr6ObPD0ic0ncGmuKVcsbUIHe 8JmVOgCxaTl2HnH3BNcvUHmsVeAL5PSyIOch80uOG7Nsh7OjZrQ8fAFqITGHzKy9 wb+C/pO1wF8xj+W9BJ5eYlvUYJLx/t/PZxFExrqrAGUxwlsqDqzoPbQsvvDAW1fB JAMUCR/mLWTQhILOYd1Dvt8gw5uFaiyqqR12f3IOC7yh3xVo3qR3CKFneH4LzuXG 7h7Y8DICXG5xm++kwjBKDbNB+p0o2Z4D56TxdNETxO2vzjnrkIspNhrMpWucvWo3 Wi6qM56W9/2KSupvdpbaOpE0rS1QZxS8v04tpB19c85omCJRDTLlu7Yark7iwo9t M+GZ0sS8CnqD1delky4T/JBT0tWMqGncXX9fWs0FNIjwOtzVY1QnsGmoEl7Sr2wG jPZRrSr2gPoDE9J3+fvGceZF9qEPe9+PDA7boetokxvVS6+mFI3IsPc3wC50uHsz 27gnUH0WovUT7iIcxj/8NACejaaPCR6To9ozfhJbsLEh6YzQc+rQajw/Ag3jr933 LuyEK93PYAVQ2391aPBIDeBcDO0oKvy71foWV3sTTVyu7kKnPLRxtVsXbRv6cfQX 1f5O4GM8SzT1HuNvwMS6 =z5IY -----END PGP SIGNATURE-----

3 2