#extent-root implies support for #extent-auto

[skynavga]

IMSC1.0 specifies that #extent-root is permitted, and #extent-root implies support for #extent-auto (see note in TTML2 §E.1.92); however, IMSC1.1 does not include #extent-auto in §6, which would mean that #extent-auto is prohibited in IMSC1.1. This makes IMSC1.1 incompatible with both IMSC1.0 and TTML2 regarding the semantics of #extent-root.

[palemieux]

This makes IMSC1.1 incompatible with both IMSC1.0

IMSC 1.0 specifies that the tts:extent attribute SHALL be present on all region elements, where it SHALL use px units or "percentage" syntax, thereby effectively prohibiting #extent-auto.

[skynavga]

The language is ambiguous since px units and percentage syntax only apply to length expressions. To remove the ambiguity, the language should be changed to read

The tts:extent attribute SHALL be present on all region elements, where it SHALL use <length> expressions that specify px units or "percentage" syntax.

[palemieux]

@skynavga Did you consider the prose in the latest Editor's Draft?

[skynavga]

ok, the language is fine in 1.2

[nigelmegitt]

Can we close this issue then @skynavga?

[palemieux]

Summary:

8.4.2 #extent-region
The tts:extent attribute SHALL be present on all region elements, where it SHALL use px units, percentage values, or root container relative units.

is replaced with:

8.4.2 #extent-region
For each region element defined in the Document Instance, the specified value of the tts:extent style property SHALL consist of two length expressions
that use pixel (px), percentage (%), or root container relative units.

[css-meeting-bot]

The Timed Text Working Group just discussed #extent-root implies support for #extent-auto imsc#489, and agreed to the following:

• SUMMARY: Reconsider the wording for this in conjunction with IMSC 1.2 #475 and come up with something all are happy with.
• RESOLUTION: Approve this erratum as summarised at https://github.com/w3c/imsc/issues/489#issuecomment-562895235

The full IRC log of that discussion

<nigel> Topic: #extent-root implies support for #extent-auto imsc#489
<nigel> github: https://github.com//issues/489
<nigel> Pierre: I copied the text from IMSC 1.2 which we had agreed to.
<nigel> Cyril: I'm trying to understand what it means - is it specifying format if a value is specified or is it requiring a value?
<nigel> Pierre: "specified value" has a defined meaning in TTML
<nigel> Cyril: I understand, I'm asking if the attribute is required on the element or not?
<nigel> Pierre: The only way to have a specified value is if it is specified.
<nigel> Cyril: Sure, but this text doesn't say what happens if it is not specified.
<nigel> Pierre: If the author does not specify a value then the specified value is undefined, so it cannot be.
<nigel> Cyril: It does not exist...
<nigel> Pierre: Exactly, so it cannot be a length expression, therefore it implies it is required.
<nigel> .. The implication is the only way to satisfy the constraint is to include the attribute.
<nigel> .. We should fix this in IMSC 1.2 and then port it back.
<nigel> Nigel: We can agree the words here and do it here and in IMSC 1.2.
<nigel> Pierre: Absolutely. We can hold off making this change now and come back to it next meeting in case there is a better idea.
<nigel> Nigel: Simple wording change - add "is required to be present"
<nigel> Pierre: Folk weren't happy with that previously on the thread.
<nigel> Nigel: I don't see that here - Glenn's comment on 10th Oct included it for example.
<nigel> Pierre: I'll point to the IMSC 1.2 issue that was closed on this.
<nigel> .. It is #475
<nigel> .. It's a long thread. Suggest Cyril reopens the issue and adds the suggestion.
<nigel> Cyril: Ok will do.
<nigel> SUMMARY: Reconsider the wording for this in conjunction with IMSC 1.2 #475 and come up with something all are happy with.
<nigel> Pierre: Remember we have to be careful about `<set>`, initial value etc.
<nigel> .. Specify does not necessarily mean it is on the element itself, which is why "is present" is not awesome.
<nigel> Cyril: Being consistent with oneself is difficult!
<nigel> Pierre: Thankfully we have GitHub to remind ourselves.
<nigel> Nigel: We've toyed with trying to work "computed" in here in the past too.
<nigel> Pierre: Yes, and the reason it was open is that style properties can be specified using child style elements, which is
<nigel> .. the equivalent of specifying the style property on the element itself.
<nigel> .. There are many ways for something to be considered specified.
<nigel> Nigel: I think we're talking about the value of the attribute in the specified style set for the region element, following
<nigel> .. the style resolution process.
<nigel> Pierre: That term in TTML2 is "specified style"
<nigel> .. What I'm saying is "shall be present" is not right because it excludes the example that Cyril raised in #475.
<nigel> Nigel: In TTML2 terminology section "specified style set" is defined but not "specified style". It may be elsewhere.
<cyril> https://developer.mozilla.org/en-US/docs/Web/CSS/specified_value
<nigel> Cyril: It's an XML term?
<nigel> Glenn: I would avoid using CSS terminology.
<nigel> Pierre: 10.4.3.1 in TTML2
<nigel> -> https://www.w3.org/TR/ttml2/#semantics-style-resolved-value-category-specified TTML2 10.4.3.1 Specified Values
<nigel> Cyril: Ok can we link to that?
<nigel> .. I will change my comment.
<nigel> Pierre: It is actually already there in IMSC 1.2 - specified value already says that under #extent-region.
<nigel> .. Your wish has already come through.
<nigel> Cyril: Apologies, I'll delete my comment and close the issue.
<nigel> Pierre: Going back to IMSC 1.1 do we have to add this link?
<nigel> Nigel: I think we should
<nigel> Pierre: I will modify that then.
<nigel> .. [adds extra text to signify the meaning of "specified value"]
<nigel> .. done
<nigel> Nigel: I see that "specified value" is now a link.
<nigel> Pierre: Shall we approve this erratum?
<nigel> Nigel: Any objections to approving this erratum?
<nigel> group: [no objections]
<nigel> RESOLUTION: Approve this erratum as summarised at https://github.com//issues/489#issuecomment-562895235
<nigel> Pierre: The text will be taken from the summary, I understand.