Re: Verbosity ot wpt-stability-bot

On Thu, Dec 1, 2016 at 10:50 PM James Graham <james@hoppipolla.co.uk> wrote:

> On 30/11/16 19:01, Philip Jägenstedt wrote:
> > Hi all,
> >
> > I asked James about this and he asked me to seek feedback here.
> >
> > The wpt-stability-bot is amazing, but I wonder if there are any options
> for
> > making it generate less email. On
> > https://github.com/w3c/web-platform-tests/pull/4250 I did a lot of
> fixups
> > while experimenting, and wpt-stability-bot has posted a total of 18
> > comments, obscuring any real discussion.
> >
> > I think seeing the pass/fail results is useful even if it's stable, so
> some
> > visibility is great. How about it the bot only added one comment, and
> kept
> > updating that with the most recent info?
>
> Done.
>
> Note that using the details element for the results, as has also been
> suggested, is non-trivial, because:
>
>   * The GitHub stylesheet breaks the disclosure triangle in
> spec-conformant browsers
>   * It isn't possible to add inline style attributes
>   * It isn't possible to have markdown inside an explicit tag, so the
> tables would need to be rewritten using explicit <table>+etc. elements
>   * The current code logs exactly the same plaintext to the travis log
> and the github comment, using the fact that markdown is more or less
> readable as text. Therefore replacing all the markdown with explicit
> tags is highly undesirable since human readability of the logs is
> essential when PRs are not submitted to w3c/web-platform-tests but a fork.
>

Thank you James, that is great! I see there's still a separate comment for
Chrome and Firefox, but I guess that's intended? I'll see if in my next
fixup rampage if it's really working :)

Received on Tuesday, 6 December 2016 15:34:42 UTC