Identify sections at risk for 5.2

[chaals]

As preparation for the HTML 5.2 Candidate Recommendation, we should identify anything in the specification that does not have interoperable implementation. Roughly, this means at least two working implementations that are not marked for deprecation - and ideally, some sense that further implementations are likely. Please list candidate sections in this issue, or open an issue suggesting something be marked at risk and noting why...

[chaals]

Proposed: inputmode.

According to CanIUse it was implemented by Firefox, and it has been implemented repeatedly in the past by mobile browsers, although I do not know of interop testing. Note that "autocapitalize" has been suggested as having implementation: #208

[chaals]

Proposed: menu and menuitem

The only known implementation is Firefox. While I know of claims that Blink implemented this, it appears not to work in any blink-based browser, even with the enabling flag set.

[chaals]

proposed: allow-presentation as a sandboxing flag for iframes - see #437

[chaals]

Proposed: dialog
It's implemented in blink. There is an implementation behind a flag in Firefox, but it's pretty flaky. If it got more work, then this feature might be ready to land. Otherwise, I think not yet...

[chaals]

Proposed: input type="datetime"

Still doesn't seem to be supported, running a simple test

[chaals]

Proposed: The plugin API

Browsers are consistently tightening the constraints on this, to the point where it seems unlikely to continue being interoperable.

We should examine this and if appropriate remove it.

[prlbr]

<menuitem> has been removed from Blink: https://bugs.chromium.org/p/chromium/issues/detail?id=87553&desc=2#c66 (via whatwg/html#2730)

[siusin]

The at-risk features for HTML5.2 have been announced at the latest WD.

[chaals]

@tidoust do you have implementation info for allow-presentation ?

[chaals]

<input type="datetime"> doesn't work in recent Blink, Edge or Gecko. As far as I know it is not implemented in Webkit either. Suggest removal

<menuitem> works in Firefox but not Edge nor Yandex (Blink). As far as I know it is not supported in Webkit but I didn't test that. Suggest removal.

CaniUse says inputmode doesn't work and I don't know of any modern implementation still shipping. Suggest removal.

[chaals]

The chairs/editors propose to keep dialog given that there are two implementations...

[chaals]

The Plugins API has been moved to obsolete, rather than removed entirely.

[chaals]

allowpresentation as a flag on iframe should be removed from 5.2 (but not 5.3)

[chaals]

This is done...

[csarven]

@chaals re #893 (comment) , on what grounds did menu made its way into the spec?

[LJWatson]

@csarven it didn't. <menu> was pulled from HTML5.2

[csarven]

@LJWatson I'm aware (hence this issue and #893 (comment) I'm referring to). It was removed on the grounds that only Firefox supports it. I'm asking on what grounds was it introduced to 5.x to begin with.

[chaals]

@csarven I think it seemed like a good idea at the time, to those who were working on it. There were other implementations to the point where we didn't pull it from 5.1, and even in the 5.2 timeframe there was "implementor interest". Once upon a time that was considered good enough to put things in an HTML spec, before the world realised that this didn't translate into reality well.

[LJWatson]

The background is that <menu> was redefined in HTML5.0, having been present in HTML since the early days. Looking briefly through the HTML5.0 test results, it seems that all tests relating to menu passed with two (or in some cases more) implementations at the time HTML5 went to Rec.

[csarven]

@chaals Are you particularly referring to Web browser implementors as opposed to any?

@LJWatson Thanks for sharing! That's interesting.

Is there a general protocol on addressing potential backwards compatibility issues - whether within the HTML WG or in coordination with the other W3C WGs?

If/how are HTML documents in the wild, non-browser tools, and W3C specs (WAI*) referring to components in HTML - like the concept of an interactive element "menu" - should handle the changes in the HTML spec? Or is that out of scope for the HTML WG?