Browser Extension CG teleconf — 23 Mar 2017

Agenda

See also: IRC log

Attendees

Present
Florian Rivoal (Florian), Mike Pietraszak (mikepie), Scott Low (scottlow)
Regrets
Chair
Florian
Scribe
Florian

Contents


Agenda

Florian: Not having full attendance, should we work on only part of the agenda?
... Demo, manifest, tests?

mikepie: Sure.

Demo

mikepie: I wrote an extension that tests the browser.runtime API, and run it from my WPT fork https://github.com/mikepie1/web-platform-tests/tree/mikepie1/browserext/)
... This will run in FF, Edge, and Chrome. I have not tested Opera yet.
... The WPT folder structure mirrors spec sections.

Florian: Mirroring spec sections is good. Also, make sure to use the exact id used by each section, so that software can automatically match the section and corresponding tests.

mikepie makes a demo.

scottlow: Should we test webdriver itself?

Florian: I think so.

mikepie: I'll get advice from experts on how to do that.

Florian: Can you change the 2 sec wait to a 2 sec timeout, letting the test finish faster if it can? Thinking of continuous integration.

mikepie: Will do.
... After failing a test if the browser object is missing, I alias any vendor specific name to the "browser" name so that other tests can run.

Florian: This sounds useful. We should check with WPT people if they already have a policy on this.
... A key point is to run this in CI. Since extension tests are new, we need to check with each browser if they are OK with this structure for tests to get picked up by CI systems.

mikepie: I'll check on our side.

Florian: We need to check with WPT if we're doing things the right way, and who needs review/merge access, since WPT has a policy of review-then-merge.

mikepie: </demo>

Manifest

Florian: I've done a quick review of the manifest section, and overall it's good.
... There are a few place where you can guess what the spec means, but it isn't actually saying it unambiguously. Also places where it says something is required, but not what happens if you omit it.
... Will send details on github.

Florian: Also, the pre-parsing section says to strip "//" comments, but does not say anything about "/* */" comments.

mikepie: I believe we decided on that way back at the beginning of the CG. I do not remember why.

Florian: I do remember we decided to strip comments, but I have no memory of a distinction between "//" and "/**/".
... I'll check minutes to see if there was a rationale, and otherwise will send a pull request.

Tests

Florian: we covered this during the demo.

Next meeting

Florian: Given the limited attendance today, it doesn't make a whole lot of sense to decide on a date during the call. Let's see over email, trying for next week.