Re: A new GitHub organization for web-platform-tests

Often good to batch the pain, but maybe not this time, then :)

For people interested in following along closely, I've started a doc
describing the transition:
https://docs.google.com/document/d/1S_WSpZVyzIKgnFo2Oc3cgSMd46NsmO9s7-TI3rKhmWM/edit?usp=sharing

Too many details to work out to ask for review, but take a look if you're
very curious.

On Wed, Apr 4, 2018 at 5:53 PM Mike Pennisi <mike@bocoup.com> wrote:

> > What would a structure that doesn't mix tests and infrastructure look
> > like?  Two subdirectories?
>
> Due to James' feedback, it probably doesn't matter. But since you ask:
> we have a bunch of non-test directories that could remain at the top
> level or be re-organized. The latter would be more disruptive, of
> course, but this suggestion is couched by, "if we're already asking
> consumers to accommodate breaking changes..."
>
> > I don't think there's a reason to link together the proposal to move
> > to an independent Github organization (which seems to have rough
> > consensus already), and your (rather more vague) proposal.
>
> My rationale was that since the organization change will interrupt the
> normal workflow for WPT, it may be a good opportunity to include
> additional changes that we've previously deferred out of reluctance to
> disturb consumers.
>
> > Any mass move like this would be disruptive for gecko.
>
> Thanks, James. This is reason enough to drop the proposal. Just wanted
> to make sure we weren't missing out on an opportunity to improve WPT
> still further :)
>
> On 04/04/2018 05:31 AM, James Graham wrote:
> > On 04/04/2018 07:18, Philip Jägenstedt wrote:
> >> What would a structure that doesn't mix tests and infrastructure look
> >> like?
> >> Two subdirectories?
> >
> > Any mass move like this would be disruptive for gecko. Although we do
> > have support for moving metadata files when downstreaming, I don't
> > particularly want to find out all the edge cases that occur when every
> > test is moved. Moving the tooling is not well supported since various
> > paths are hardcoded (so that e.g. wptrunner ends up on the Python path
> > when running gecko-specific tooling). It would also cause merge
> > conflicts with every local change, and break tests relying on absolute
> > paths. I can't imagine it would be less than a week of work to fix all
> > the fallout from such a change, and more than likely there would be
> > subtle breakage not noticed for a long time.
> >
> > In the absence of strong evidence that the current setup is causing
> > problems at the same scale as the disruption any move would cause, I
> > am very reluctant to start making large-scale changes to the
> > organisation of the repository.
> >
>
>
>
>