Re: Porting wptserve handlers to Python 3: next steps / code review

Attempting to answer the question for Stephen since I've poked a bit at
wpt-pr-bot in the past.

In
https://github.com/web-platform-tests/wpt-pr-bot/blob/master/lib/comment.js we
have access to the inferred labels (as metadata.labels) and could use them,
but the issue is that the python3 label is now added by hand. One could try
to add support for acting on labels that are added at PR creation time, or
reaction to added labels, but it the resulting behavior would not be easy
to understand, I predict.

I guess there's no way to infer from a PRs content alone if it's a python3
conversion?

On Fri, May 15, 2020 at 7:41 PM Robert Ma <robertma@chromium.org> wrote:

> Thank you, Josh & Philip!
>
> I've updated the RFC <https://github.com/web-platform-tests/rfcs/pull/49> to
> reflect our latest consensus (preferring bytes everywhere). Please take
> another look (the diff
> <https://github.com/web-platform-tests/rfcs/pull/49/commits/2e95208da921d31cfbd38a37204777bd62ef1fd1>
> isn't big because I basically swapped two sections with some additions).
>
> Stephen, would it be possible to tweak wpt-pr-bot to handle PRs with the
> "python3" label and assign them to a special pool of reviewers?
>
> On Wed, May 13, 2020 at 8:17 AM Philip Jägenstedt <foolip@chromium.org>
> wrote:
>
>> I can confirm I volunteer to review!
>>
>> On Tue, May 12, 2020 at 8:54 PM Josh Matthews <josh@joshmatthews.net>
>> wrote:
>>
>>> I've got some experience porting the eventsource handlers (although I
>>> haven't submitted that PR yet due to a couple unfinished tricky handlers),
>>> so I can review PRs as well.
>>>
>>> Cheers,
>>> Josh
>>>
>>> On Tue, 12 May 2020 at 13:03, Stephen Mcgruer <smcgruer@chromium.org>
>>> wrote:
>>>
>>>> Thanks for sending this email Robert; I'm excited to see us keep the
>>>> ball rolling on Python 3 support.
>>>>
>>>> > In addition, we'd really appreciate a few more people to sign up for
>>>> reviewing these changes to share the workload. Anyone volunteering?
>>>>
>>>> I'm happy to review PRs, albeit with no specific prior knowledge.
>>>> +foolip, who volunteered to review as well.
>>>>
>>>> That'd bring us to 5 reviewers assuming jgraham and annevk are willing
>>>> to review; do you think that is enough Robert?
>>>>
>>>> On Mon, 11 May 2020 at 18:19, Robert Ma <robertma@chromium.org> wrote:
>>>>
>>>>> Hello everyone,
>>>>>
>>>>> We're getting close to finalizing the plan for migrating close to 500
>>>>> wptserve handlers we have in WPT. Now we have a few concrete steps to take:
>>>>>
>>>>> 1. Regarding the trial PR
>>>>> <https://github.com/web-platform-tests/wpt/pull/23363>, James, Anne
>>>>> and others who'd like to take a look, do you have any other comments on
>>>>> this PR, especially high-level ones about the general approach? This would
>>>>> unblock the following steps and we can address small issues in parallel.
>>>>> 2. If we agree this approach is what we wanted by having consistent
>>>>> and explicit semantics across Python 2 and 3, I'll update the RFC
>>>>> <https://github.com/web-platform-tests/rfcs/pull/49> (essentially
>>>>> swapping the currently "recommended" and "alternative" approaches and
>>>>> filling in some more concrete guidelines), and kick off a new round of RFC
>>>>> process (hopefully relatively quick since many people are already on board
>>>>> with the new approach).
>>>>> 3. Meanwhile, Ziran can start porting more handlers (we can wait until
>>>>> the RFC is accepted to actually merge the PRs). We have hundreds of
>>>>> handlers and we should expect lots of PRs. Reviewing them is a critical
>>>>> task, too. Since we now have concrete guidelines and changes will be
>>>>> largely mechanical, I'm proposing to adopt the "LGTM % nits" convention
>>>>> widely used in Chromium: if a PR largely looks good but has some minor
>>>>> issues, approve the PR with comments. In addition, we'd really appreciate a
>>>>> few more people to sign up for reviewing these changes to share the
>>>>> workload. Anyone volunteering?
>>>>>
>>>>> Best,
>>>>> Robert
>>>>>
>>>>