Don't react to PR comments

[foolip]

This simplifies the filtering and removes the need to fetch the pull
request in getPullRequest.

[foolip]

The code changes here are probably straightforward enough to review, but the key concern here should be unintended side effects of this change:

• Could the code be broken in such a way that comments were previously the only thing that triggered the right things, given that a PR description counts as a comment?
• Is wpt-pr-bot reliable, and if not is that effectively masked by reacting to comments also?

[jugglinmike]

I like this simplification! We should also reconfigure the GitHub Webhook by removing the "Issues" and "Issue comments" events. In that case, I'd push to remove the explicit filtering (since that would be dead code, and it would insinuate the possibility of unreachable states).

• Could the code be broken in such a way that comments were previously the only thing that triggered the right things, given that a PR description counts as a comment?
• Is wpt-pr-bot reliable, and if not is that effectively masked by reacting to comments also?

I'm not aware of any problems like that, but I can't say they don't exist. We could build some confidence by inspecting "timelines" from recent pull requests, looking for places where the script acted directly after a comment.

[zcorpan]

For #33 and #81 the bot may need to listen for new comments again, unless we use labels instead to communicate with the bot.

[foolip]

Yes, that's a good point. Perhaps that could then be isolated from the code that reacts to PR creation/update, though? It was quite hard to follow the code as it was here.

[zcorpan]

Perhaps that could then be isolated from the code that reacts to PR creation/update, though?

It probably could, yeah.

[jugglinmike]

Making this script readable and testable is a worthy goal, especially if we're intending to continue extending it. I think that will require a more dedicated effort, though. It'd be great to bite the bullet for the technical debt and move forward with confidence.

[foolip]

I still see #77 as plan A, certainly for adding labels and reviewers. @zcorpan's upcoming work on improving the review process might reveal some needs that we can't do with GitHub Actions and maybe we'll have to keep operating this bot, but most of the code that exists I'm not confident will continue to live in this repo. That is to say, improving as we make changes is a good idea, but wrangling it all under control could well turn out to be a wasted effort.

[foolip]

I believe this will fix #69. But it might be a good idea to merge #118 and add some minimal test coverage for index.js first.

[foolip]

@zcorpan per discussion with @boazsender I'm reassigning this and two other of my PRs to you. Not super high priority but would be nice to get it done within a few weeks.

[zcorpan]

In web-platform-tests/wpt#20476 the comment produced this in the papertrail

Nov 27 13:33:51 wpt-pr-bot app/web.1 Ignoring event: not a pull request

[foolip]

@zcorpan sweet, thanks! Does everything still seem to be working for a new and updated PR?

[zcorpan]

Yes, I think so. I haven't tested updating PRs, but the papertrail shows recent activity for e.g. web-platform-tests/wpt#20441