Thread: Upgrading to newer Django? Reason for including django-selectablelibrary in repo?

Hi Kalvin,

On 3/3/20 12:57 PM, Kalvin Eng wrote:
> Hi there,
>
> I'm new to the community and looking to send a GSoC proposal for the
> performance farm project
> (https://wiki.postgresql.org/wiki/GSoC_2020#Develop_Performance_Farm_Benchmarks_and_Website_.282020.29).

Cool :)

> One of the requirements was integrating the performance farm app with
> pgweb to include already registered users.
> So I had a look at the code for pgweb to understand the login flow for
> integration.
>
> Upon looking at the code for pgweb, I'm wondering if these things should
> be patched:
>
>   * Upgrading to Django 2.2 LTS since Django 1.11 will no longer be
>     supported in April 2020

Seems like a good idea. IIRC Magnus did some testing way back on 2.0
(unless I'm imagining)

>   * Remove django-selectable from the repo as it is available in pypi
>     (https://pypi.org/project/django-selectable/)
>
>
> Thanks,
> Kalvin

(Take 2 without prematurely sending. This is why I don't have nice things)

On 3/3/20 5:10 PM, Jonathan S. Katz wrote:
> Hi Kalvin,
>
> On 3/3/20 12:57 PM, Kalvin Eng wrote:
>> Hi there,
>>
>> I'm new to the community and looking to send a GSoC proposal for the
>> performance farm project
>> (https://wiki.postgresql.org/wiki/GSoC_2020#Develop_Performance_Farm_Benchmarks_and_Website_.282020.29).
>
> Cool :)
>
>> One of the requirements was integrating the performance farm app with
>> pgweb to include already registered users.
>> So I had a look at the code for pgweb to understand the login flow for
>> integration.
>>
>> Upon looking at the code for pgweb, I'm wondering if these things should
>> be patched:
>>
>>   * Upgrading to Django 2.2 LTS since Django 1.11 will no longer be
>>     supported in April 2020
>
> Seems like a good idea. IIRC Magnus did some testing way back on 2.0
> (unless I'm imagining)

*I believe Magnus had tested it on 2.0 and it either mostly or all
worked. I don't think we've done too much wonky development on the
backend since the Python 3 move. Needs some thorough testing, of course.

>>   * Remove django-selectable from the repo as it is available in pypi
>>     (https://pypi.org/project/django-selectable/)

In production we use the system packages, so we'd have to ensure that
it's there first before removing it (which otherwise is why it'd be an
embedded dependency)

Thanks,

Jonathan

On Tue, Mar 3, 2020 at 2:16 PM Jonathan S. Katz <jkatz@postgresql.org> wrote:
>
> (Take 2 without prematurely sending. This is why I don't have nice things)
>
> On 3/3/20 5:10 PM, Jonathan S. Katz wrote:
> > Hi Kalvin,
> >
> > On 3/3/20 12:57 PM, Kalvin Eng wrote:
> >> Hi there,
> >>
> >> I'm new to the community and looking to send a GSoC proposal for the
> >> performance farm project
> >> (https://wiki.postgresql.org/wiki/GSoC_2020#Develop_Performance_Farm_Benchmarks_and_Website_.282020.29).
> >
> > Cool :)
> >
> >> One of the requirements was integrating the performance farm app with
> >> pgweb to include already registered users.
> >> So I had a look at the code for pgweb to understand the login flow for
> >> integration.
> >>
> >> Upon looking at the code for pgweb, I'm wondering if these things should
> >> be patched:
> >>
> >>   * Upgrading to Django 2.2 LTS since Django 1.11 will no longer be
> >>     supported in April 2020
> >
> > Seems like a good idea. IIRC Magnus did some testing way back on 2.0
> > (unless I'm imagining)
>
> *I believe Magnus had tested it on 2.0 and it either mostly or all
> worked. I don't think we've done too much wonky development on the
> backend since the Python 3 move. Needs some thorough testing, of course.

I don't think I said that, but I prorbably said that we *should* do that :)

That said, yes, it's 2.2 that we should be upgrading to, and not 2.0.
And the sooner we can look at that, the better. We also have a bunch
of other code that needs to get update to that, so if we can get pgweb
dealt with fairly early and can copy patterns ou tof that it would
definitley be helpful for a number of other use cases as well.


> >>   * Remove django-selectable from the repo as it is available in pypi
> >>     (https://pypi.org/project/django-selectable/)
>
> In production we use the system packages, so we'd have to ensure that
> it's there first before removing it (which otherwise is why it'd be an
> embedded dependency)

It's not available, and that's one of the main reasons it got vendored
inthe first place. There was some other reason as well,  We have since
changed exactly how we deploy django centrally so it might be that
it's a lot easier to fix now, so that's somethign we should
investigate. But it's definitely not as easy as just removing it and
adding it to requirements.txt.

For some of the other projects we have that use it, the plan has been
to just get rid of it instead of investigating that part. But given
that pgweb still relies fairly heavily on the django admin for things
that now have a non-trivial amount of entries in some dropdowns, we
can probably not just remove it without having done a lot of other
things first -- so finding a better deployment model for it is
definitely a good idea.

-- 
 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/