1270459 - .scope property on td/th should be limited to only known values

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Aryeh Gregor (:ayg) (no longer with Mozilla)]

Spec:

"The scope IDL attribute must reflect the content attribute of the same name, limited to only known values."
https://html.spec.whatwg.org/multipage/tables.html#dom-th-scope

Existing UAs do not limit to known values (tested in Firefox, Chrome, and IE 11).  I recall this being discussed somewhere (probably on GitHub, but I can't find it), and the conclusion was that we'd prefer to make enum properties reflect known values if we can get away with it.  From what I was told, the usage of .scope is about zero, so it shouldn't hurt to limit it.

Comment 1

[Aryeh Gregor (:ayg) (no longer with Mozilla)]

Attachment: 0001-Bug-1270459-Limit-td-th.scope-to-known-values.patch

https://treeherder.mozilla.org/#/jobs?repo=try&revision=959ae4487347&exclusion_profile=false

Comment 2

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 8749150 [details] [diff] [review]
0001-Bug-1270459-Limit-td-th.scope-to-known-values.patch

r=me, I guess.  Do other UAs do this?

Comment 3

[Aryeh Gregor (:ayg) (no longer with Mozilla)]

No, as noted in comment #0.

Comment 4

[Boris Zbarsky [:bzbarsky]]

OK, then I'm a bit worried about compat impact.  I agree that the usage should be low, but still...

It might be a good idea to add some telemetry to see how much this getter is being called.

Comment 5

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/e5019fbf51a1

Comment 6

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/e5019fbf51a1

Comment 7

[Olli Pettay [:smaug][[e-mail]]]

Wouldn't it make more sense to fix the spec to follow what implementations do? 
If the usage is low, is there really any reason to change the behavior?

Comment 8

[Aryeh Gregor (:ayg) (no longer with Mozilla)]

I discussed it at some point with somebody (I forget where :( ) and was told that if there's not much compat risk or implementation burden, we'd prefer to try making the platform slightly more consistent instead of matching implementations.  I don't feel strongly either way, but since it was a simple change and I don't expect it makes any real difference to our users, I thought it was a more productive use of everyone's time to just implement the spec and be done with it than to get into a discussion.