Mailman 3 August 2018 - cython-devel

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Sept. 17, 2018

Sept. 17, 2018

On 2018-08-19 08:26, Stefan Behnel wrote: > Should we make that a new directive rather than a language level? Like "py2_str=str"? That would allow its use together with language_level=3 already in the next release. With a new new directive, you also run into compatibility problems. What should the default be? py2_str=str or py2_str=unicode? The former breaks code assuming that it's unicode and the latter doesn't really solve anything: stuff will still break when language_level=3 becomes the default. My proposal is a new setting language_level=3str (meaning: everything that language_level=3 does, except unicode_literals) and make that the default. That way, you keep full compatibility with code already setting the language_level. You also have reasonably good chances that code that currently uses the implicit language_level=2 will continue to work with language_level=3str.

3 2

OpenMP 4.5 array reductions
by Nathan Goldbaum Sept. 2, 2018

Sept. 2, 2018

Hi all, I'm curious if there would be any interest in adding support for OpenMP 4.5 array reduction in the cython compiler or alternatively detecting these cases and raising a cython compiler error. Currently cython is generating code that will compile but might lead to race conditions. See: https://github.com/cython/cython/issues/2316 https://github.com/cython/cython/issues/1504 The trouble with fixing this in the cython compiler is that adding the appropriate OpenMP pragmas might generate code that will no longer compile on compilers that don't support OpenMP 4.5. However perhaps that's a better alternative than the status quo, which is generating code that might produce random results. I'd very much appreciate any feedback or advice here as this is currently blocking our ability to easily add OpenMP to our cython code in places where we'd like threads to do parallel reductions on large arrays. I would also not be surprised if there is code in the wild that is racy and silently producing incorrect results. -Nathan

2 1

Preparing Cython 3.0 with breaking changes
by Stefan Behnel Aug. 20, 2018

Aug. 20, 2018

Hi all, following the versioning discussion, I created a milestone to collect (breaking) changes that should go into the future Cython 3.0 release. https://github.com/cython/cython/milestone/58 While a major version change is a good time to fix things that, in retrospect, have led us too far in the wrong direction, we are not planning to remove the support for existing code any time soon that relies on the current semantics. For now, it looks like the major breaking changes will be to directive defaults, which means that setting them explicitly to their current configuration (e.g. in setup.py) should keep code working across Cython releases. Feedback welcome. Stefan

2 1

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Aug. 19, 2018

Aug. 19, 2018

>> Basically I am asking for >> >> type("foo") is str >> >> both on Python 2 and Python 3. > > That's another breaking change I'm not following here... what I'm asking is how Cython behaves currently (with language_level=2).

2 1

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Aug. 18, 2018

Aug. 18, 2018

On 2018-08-18 10:52, Robert Bradshaw wrote: > Do the existing c_string_type and c_string_encoding directives not > cover this usecase? Those seem to refer to C strings, I am talking about the Python type of string literals. > I suppose what you're asking for is str being the > str of the runtime, even if language_level=3 is set. Basically I am asking for type("foo") is str both on Python 2 and Python 3.

2 1

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Aug. 18, 2018

Aug. 18, 2018

On 2018-08-17 11:44, Stefan Behnel wrote: > Hej, that gives us an alternative for the versioning switch. We could > release Cython 3.0 when we change the default language level (and require > users to select "language_level=2" for legacy code). Definitely a breaking > change that merits inceasing the major version, and 3.0 seems very suitable. One annoying point with language_level=3 is that all string literals become unicode (like from __future__ import unicode_literals). Unlike the other changes that language_level=3 makes, this is a major breaking change on Python 2. So I would very much prefer enabling everything that language_level=3 does, but keeping strings of type "str" on Python 2. Maybe you could invent a new option for that, say "language_level=3str"? Jeroen.

2 1

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Aug. 18, 2018

Aug. 18, 2018

On 2018-08-17 11:44, Stefan Behnel wrote: > https://github.com/cython/cython/issues?q=is%3Aissue+is%3Aopen+label%3A%22P… Please add https://github.com/cython/cython/issues/1635

1 0

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Aug. 18, 2018

Aug. 18, 2018

On 2018-08-17 21:51, Robert Bradshaw wrote: > It'd be nice to change the default of cython.binding as well (breaking > change). Ideally, after PEP 580 is accepted :-)

1 0

Re: [Cython] Cython 0.29 – or 29.0 ?
by Jeroen Demeyer Aug. 17, 2018

Aug. 17, 2018

I vote for 1.0 Version 29.0 sounds too much like marketing.

8 11

Cython 0.29 – or 29.0 ?
by Stefan Behnel Aug. 15, 2018

Aug. 15, 2018

Hello everyone, I had a request during my talk at EuroPython this year to help the aspiring Cython users out there convince their managers to accept Cython in their tech stack. Cython has been around for more than a decade now, has been in heavy production use pretty much all of that time, has proven to have everything it needs as a programming language, and still resorts to a 0.x versioning scheme. I think it's time to leave the 0.x behind us and make it clear that it's a reliable programming language that won't easily change in a major code breaking way anymore. Which version should we choose next? Some projects have adopted year based release schemes, e.g. pip recently jumped from 10.x to 18.0. Browsers follow the marketing path and boldly increase their version number by a whole 1 on every feature release. I'd be happy to follow them and release a Cython 29.0 next, mostly to make it clear that we're not really changing anything from the previous release cycles, but really just move the versioning scheme one place to the left to get rid of the uninformative "0." prefix. What do you think? Stefan

2 1