Re: [Cython] Preparations for Cython 0.26.1 and 0.27
One more thing: If there was a nightly wheel like for scipy and numpy, we could run CI with Cython master. That might lead to finding issues earlier. We probably don't want to build cython from source on CI, as I expect that would take too much time. On 08/02/2017 05:44 AM, Stefan Behnel wrote: Hi! Given the current set of bugs and regressions in 0.26, I think we should start preparing a 0.26.1 release. I've already updated the release branch with recent fixes. It's currently the same as master, but I'd like to separate the branches soon in order to start merging features for 0.27 to avoid bit-rot. The following regressions seem worth looking into for 0.26.1: https://github.com/cython/cython/issues/1790 https://github.com/cython/cython/issues/1788 Anything else of importance? Stefan ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
Hi, On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote: > One more thing: If there was a nightly wheel like for scipy and numpy, > we could run CI with Cython master. That might lead to finding issues > earlier. > We probably don't want to build cython from source on CI, as I expect that > would > take too much time. That's a good idea - I will set that up, and email back here with details. Cheers, Matthew ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
On Thu, Aug 3, 2017 at 10:05 AM Matthew Brett wrote: > Hi, > > On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote: > > One more thing: If there was a nightly wheel like for scipy and numpy, > > we could run CI with Cython master. That might lead to finding issues > > earlier. > > We probably don't want to build cython from source on CI, as I expect > that > > would > > take too much time. > > That's a good idea - I will set that up, and email back here with details. > > Cheers, > > Matthew > ___ > cython-devel mailing list > cython-devel@python.org > https://mail.python.org/mailman/listinfo/cython-devel FWIW, depending on your CI needs using the no-cython-compile option can speed things up pretty drastically. Cython files don't compile quite as quickly with that option set, but installing Cython itself is much faster. Best of luck, Ian Henriksen ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
Ian Henriksen schrieb am 03.08.2017 um 18:16: > On Thu, Aug 3, 2017 at 10:05 AM Matthew Brett wrote: >> On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote: >>> One more thing: If there was a nightly wheel like for scipy and numpy, >>> we could run CI with Cython master. That might lead to finding issues >>> earlier. >>> We probably don't want to build cython from source on CI, as I expect >>> that would take too much time. >> >> That's a good idea - I will set that up, and email back here with details. > > FWIW, depending on your CI needs using the no-cython-compile option can > speed > things up pretty drastically. Cython files don't compile quite as quickly > with that option set, but installing Cython itself is much faster. Yes, for one-shot compilations that has always been the recommended way, if there really is no suitable binary wheel available. Cython can always be installed and used as a pure Python package, and even run directly from a git checkout (or submodule if you really feel like it), without any installation, just by adding it to the PYTHONPATH. Stefan ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
On 08/03/2017 12:46 PM, Stefan Behnel wrote: Ian Henriksen schrieb am 03.08.2017 um 18:16: On Thu, Aug 3, 2017 at 10:05 AM Matthew Brett wrote: On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote: One more thing: If there was a nightly wheel like for scipy and numpy, we could run CI with Cython master. That might lead to finding issues earlier. We probably don't want to build cython from source on CI, as I expect that would take too much time. That's a good idea - I will set that up, and email back here with details. FWIW, depending on your CI needs using the no-cython-compile option can speed things up pretty drastically. Cython files don't compile quite as quickly with that option set, but installing Cython itself is much faster. Yes, for one-shot compilations that has always been the recommended way, if there really is no suitable binary wheel available. Cython can always be installed and used as a pure Python package, and even run directly from a git checkout (or submodule if you really feel like it), without any installation, just by adding it to the PYTHONPATH. Thanks, we'll give that a go. We have a relative substantial amount of cython code, so let's see how that'll impact runtime. ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
Hi, On Thu, Aug 3, 2017 at 4:04 PM, Matthew Brett wrote: > Hi, > > On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote: >> One more thing: If there was a nightly wheel like for scipy and numpy, >> we could run CI with Cython master. That might lead to finding issues >> earlier. >> We probably don't want to build cython from source on CI, as I expect that >> would >> take too much time. > > That's a good idea - I will set that up, and email back here with details. OK - I have set that up, wheels building on the "daily" branch here: https://travis-ci.org/MacPython/cython-wheels/branches But, I'm a bit worried by the pre-release naming scheme, because the built wheel has name of form "Cython-0.26.1a0-cp..." . Can I ask - where does the "1a0" come from? Can it be used to select the most recent pre-release wheel? I mean, if we build a wheel from a later commit, will pip find the later wheel before this one? Cheers, Matthew ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
Hi,
On Thu, Aug 3, 2017 at 2:39 PM, James C. McPherson
wrote:
> On 4/08/17 04:10 am, Matthew Brett wrote:
>>
>> Hi,
>>
>> On Thu, Aug 3, 2017 at 4:04 PM, Matthew Brett
>> wrote:
>>>
>>> Hi,
>>>
>>> On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote:
One more thing: If there was a nightly wheel like for scipy and numpy,
we could run CI with Cython master. That might lead to finding issues
earlier.
We probably don't want to build cython from source on CI, as I expect
that
would
take too much time.
>>>
>>>
>>> That's a good idea - I will set that up, and email back here with
>>> details.
>>
>>
>> OK - I have set that up, wheels building on the "daily" branch here:
>>
>> https://travis-ci.org/MacPython/cython-wheels/branches
>>
>> But, I'm a bit worried by the pre-release naming scheme, because the
>> built wheel has name of form "Cython-0.26.1a0-cp..." . Can I ask -
>> where does the "1a0" come from? Can it be used to select the most
>> recent pre-release wheel? I mean, if we build a wheel from a later
>> commit, will pip find the later wheel before this one?
>
>
> I'm pretty sure that the '1a0' comes from changeset hash,
> to reflect however many changes this build has pulled in
> which are ahead of or behind trunk.
Sorry, I'm not quite sure what you mean by changeset hash - you mean a
hash of the output of "git diff 0.26..." or something? In which case,
the output would be useless for identifying the latest wheel, which
was my worry.
> Also, from reading
> https://travis-ci.org/MacPython/cython-wheels/jobs/255434019
> I'm not entirely sure that it's all setup correctly, given
> the prevalence of messages such as these:
>
>
>
> ERROR:root:code for hash blake2b was not found.
> Traceback (most recent call last):
> File "/opt/cp36m/lib/python3.6/hashlib.py", line 243, in
> globals()[__func_name] = __get_hash(__func_name)
> File "/opt/cp36m/lib/python3.6/hashlib.py", line 119, in
> __get_openssl_constructor
> return __get_builtin_constructor(name)
> File "/opt/cp36m/lib/python3.6/hashlib.py", line 113, in
> __get_builtin_constructor
> raise ValueError('unsupported hash type ' + name)
> ValueError: unsupported hash type blake2b
Those are benign, surprisingly ... It's a little bit difficult to
build Python 3.6 with that hash algorithm, but errors don't seem to
affect the wheel builds.
Cheers,
Matthew
___
cython-devel mailing list
cython-devel@python.org
https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
On 4/08/17 04:10 am, Matthew Brett wrote:
Hi,
On Thu, Aug 3, 2017 at 4:04 PM, Matthew Brett wrote:
Hi,
On Thu, Aug 3, 2017 at 2:12 PM, Andy wrote:
One more thing: If there was a nightly wheel like for scipy and numpy,
we could run CI with Cython master. That might lead to finding issues
earlier.
We probably don't want to build cython from source on CI, as I expect that
would
take too much time.
That's a good idea - I will set that up, and email back here with details.
OK - I have set that up, wheels building on the "daily" branch here:
https://travis-ci.org/MacPython/cython-wheels/branches
But, I'm a bit worried by the pre-release naming scheme, because the
built wheel has name of form "Cython-0.26.1a0-cp..." . Can I ask -
where does the "1a0" come from? Can it be used to select the most
recent pre-release wheel? I mean, if we build a wheel from a later
commit, will pip find the later wheel before this one?
I'm pretty sure that the '1a0' comes from changeset hash,
to reflect however many changes this build has pulled in
which are ahead of or behind trunk.
Also, from reading
https://travis-ci.org/MacPython/cython-wheels/jobs/255434019
I'm not entirely sure that it's all setup correctly, given
the prevalence of messages such as these:
ERROR:root:code for hash blake2b was not found.
Traceback (most recent call last):
File "/opt/cp36m/lib/python3.6/hashlib.py", line 243, in
globals()[__func_name] = __get_hash(__func_name)
File "/opt/cp36m/lib/python3.6/hashlib.py", line 119, in
__get_openssl_constructor
return __get_builtin_constructor(name)
File "/opt/cp36m/lib/python3.6/hashlib.py", line 113, in
__get_builtin_constructor
raise ValueError('unsupported hash type ' + name)
ValueError: unsupported hash type blake2b
James C. McPherson
--
Oracle
Systems / Solaris / Core
https://www.jmcpdotcom.com/blog
___
cython-devel mailing list
cython-devel@python.org
https://mail.python.org/mailman/listinfo/cython-devel
Re: [Cython] Preparations for Cython 0.26.1 and 0.27
Am 3. August 2017 20:10:14 MESZ schrieb Matthew Brett: >OK - I have set that up, wheels building on the "daily" branch here: > >https://travis-ci.org/MacPython/cython-wheels/branches Nice. >But, I'm a bit worried by the pre-release naming scheme, because the >built wheel has name of form "Cython-0.26.1a0-cp..." . Can I ask - >where does the "1a0" come from? Can it be used to select the most >recent pre-release wheel? No, it's just the current version in Shadow.py, 0.26.1 alpha-0. It's guaranteed to be increasing, but not for each build. > I mean, if we build a wheel from a later >commit, will pip find the later wheel before this one? There's probably a way to get at the current Travis build number. If you want sequentially increasing version numbers, you could append that to the wheel version. Or the current date, which would even be a human readable age indicator. Stefan ___ cython-devel mailing list cython-devel@python.org https://mail.python.org/mailman/listinfo/cython-devel
