Re: VFX reference platform 2022 draft

Hi,

Regarding the python version selection, here is the detailed answer I 
just got from Nick Cannon on the VFX discussion list [1]:

> From this year we are more strictly applying the policy that a version 
> must be released by September 1st to be included in the Platform for 
> the following year. The reason for this is to allow software providers 
> to have time for testing and quality assurance with the final Platform 
> before the window for publishing major releases against it.
>
> We will still consider later change requests which we will share first 
> with the Working Group, and then here on vfx-platform-discuss for 
> feedback. Only if there are no objections raised will we adopt a late 
> change into a Platform.
>
> So an October release of Python is too late for our regular schedule, 
> but it can be requested for a late change at which point we would 
> solicit the community for feedback.

At least now we know. No idea how likely a late change request would be 
to be accepted though.

Bastien

[1] https://groups.google.com/g/vfx-platform-discuss/c/bnyJ2X1SwAw

On 5/19/21 4:55 AM, Campbell Barton via Bf-committers wrote:
> I'm not too keen on tying ourselves to the VFX platform again (for
> Python versions at least).
>
> The VFX platform has a track record of not upgrading Python for
> reasons that don't make sense for Blender (availability of Python-QT
> bindings & some companies simply not wanting to upgrade).
>
> Now Blender has a track record of not following the VFX platform when
> it doesn't suit us, undermining our commitment to such a decision.
> This can't be reassuring to the kinds of stakeholders who might find
> some benefit in Blender using the VFX platform.
>
> Blender used the VFX platform for long enough I think we would know if
> it was giving us much/anything in the way of tangible benefits.
>
>
> On Wed, May 19, 2021 at 12:14 AM Bastien Montagne via Bf-committers
> <[email protected]> wrote:
>> Hi,
>>
>> The only info regarding versions availability we have is that sentence:
>>
>>> A change in policy to only allow released versions to be included in
>>> each year's VFX Reference Platform
>> This is fairly vague and unclear, since it does not specify any limit
>> date...
>>
>> I asked about why they still stick to a fairly old python version on the
>> discussion thread Brecht linked (message is pending approval currently),
>> we'll see if we get an answer.
>>
>> I would indeed much rather use a version of python that is actively
>> maintained during the whole year the VFX recommendations apply. And
>> python has a history track now of being reliable on their release dates,
>> so I do think VFX could be a bit less conservative on this tool (and go
>> back to an older version in October/November in the very unlikely case
>> that python would be unable to release 3.10 on time).
>>
>> That being said, sticking to 3.9 for 2022 is not as bad as what they did
>> for 2021, so I would not be too annoyed if that was the case either.
>>
>> Cheers,
>> Bastien
>>
>>
>> On 5/18/21 2:50 PM, Ray Molenkamp via Bf-committers wrote:
>>> imho 3.9 is the only version they could have picked, the
>>> proposed release date for 3.10 is 2021-10-04 [1]. While
>>> the VFX platform aims to finalize in august [2]. I was
>>> rather vocal last year for them putting versions on that
>>> had not been released yet (some of which got delayed well
>>> into 2021) I'm happy to see for 2022 a more conservative
>>> stance has been taken.
>>>
>>> With bugfix support for 3.9 ending on 2022-05-02 [3] that
>>> does put them in an odd situation, they'll either have
>>> to recommend a version that may not be released on time,
>>> or recommend a version that will not see bug fixes for most
>>> of the year. There seemingly is no winning solution here
>>> since the problem appears to be the somewhat short
>>> support window for python releases.
>>>
>>> As much as I picked on them last year for making very
>>> strange recommendations, I feel they struck a good balance
>>> for 2022.
>>>
>>> --Ray
>>>
>>> [1] https://www.python.org/dev/peps/pep-0619/
>>> [2] https://vfxplatform.com/about/
>>> [3] https://www.python.org/dev/peps/pep-0596/
>>>
>>> On 2021-05-18 2:47 a.m., Sybren A. Stüvel via Bf-committers wrote:
>>>> Hello,
>>>>
>>>> On Mon, 17 May 2021 at 19:54, Brecht Van Lommel via Bf-committers <
>>>> [email protected]> wrote:
>>>>
>>>>> There is a draft for the next VFX reference platform up now. Since we had
>>>>> some issues with the last one, it would be good to give feedback if
>>>>> necessary.
>>>>>
>>>> Good call.
>>>>
>>>>
>>>>> Python was upgraded to 3.9, which will still trail behind 3.10 that will be
>>>>> released this year. So the question about diverging or not will remain.
>>>>>
>>>> My preference would be to, for non-LTS releases at least, stick to versions
>>>> of Python that still receive bugfixes. In the past we've had crashes of
>>>> Blender that were due to a bug in Python. The only way to solve that was to
>>>> upgrade Python itself. This in itself is rare, but I don't remember having
>>>> such issues at all until we stuck to the old py3.7 to adhere to the VFX
>>>> Platform.
>>>>
>>>> Is there anything known about the policy of the VFX Platform when it comes
>>>> to picking which Python version to stick to? Is it always going to be
>>>> "whatever version was released a year earlier"? Or is there still an
>>>> acceleration happening after sticking to py2.7 so long, and will they
>>>> eventually be targeting the latest versions? If it's the latter I'd be fine
>>>> with following the VFX Platform and sticking to py3.9 for a while longer.
>>>> If sticking to the platform means that for a significant amount of time
>>>> we'll be on versions of Python that don't receive bugfixes any more, I'm
>>>> less positive.
>>>>
>>>> Sybren
>>>> _______________________________________________
>>>> Bf-committers mailing list
>>>> [email protected]
>>>> https://lists.blender.org/mailman/listinfo/bf-committers
>>> _______________________________________________
>>> Bf-committers mailing list
>>> [email protected]
>>> https://lists.blender.org/mailman/listinfo/bf-committers
>> _______________________________________________
>> Bf-committers mailing list
>> [email protected]
>> https://lists.blender.org/mailman/listinfo/bf-committers
>
>
> --
> - Campbell
> _______________________________________________
> Bf-committers mailing list
> [email protected]
> https://lists.blender.org/mailman/listinfo/bf-committers
_______________________________________________
Bf-committers mailing list
[email protected]
https://lists.blender.org/mailman/listinfo/bf-committers