Make "Using computed attribute values" normative

[nigelmegitt]

Closes #249.

Retain and reformat the example about non-DAPT functionality.

---

Preview | Diff

[andreastai]

Looks good to me! Thanks!

[cconcolato]

Reading this text again, I find the parts about "that element does not correspond to a Script Event" confusing. It shouldn't matter whether or not it is a Script Event. Inheritance applies. Also it gives the impression that under some circumstances, intermediate div can be ignored.
I would remove the whole sentence starting with "If". Otherwise, I would rephrase saying something like:

If a workflow decides to rewrite a DAPT document to remove nested div elements, care should be taken that if those elements had inheritable attributes like begin, the attribute values like begin on the leaf divs should be adjusted.

[nigelmegitt]

I don't like the idea of making this example specifically about rewriting documents - it's about the correct interpretation of a document, regardless of what is being done with it.

Possibly this comment should be raised as a separate issue?

I'd be happy to emphasise further (informatively) that ignoring the intermediate <div> is not an acceptable thing to do.

[cconcolato]

My suggestion was a tentative rewrite of what I thought you were trying say. The alternative was to remove the sentence. My main problem is the current emphasis on elements that do not correspond to Script Events. We should not encourage that.

[nigelmegitt]

Here's another attempt at what I was trying to say, maybe this is less confusing?

Or, for another example, the computed times of an element in a DAPT document are relative to the begin time of its parent element. If a <div> element specifies a begin attribute, then the computed times of its child <div> elements are relative to that intermediate <div> element's begin time, and so on down the hierarchy. It is important to include those intermediate <div> elements' times in the computation even if the processing target is an instance of the DAPT data model in which they have no direct equivalent; otherwise the Script Event Begin and End times would be wrong.

[nigelmegitt]

Change pushed in 33c33fe. On re-reading, I made a couple of small tweaks because I thought the concept of an "intermediate" div wasn't really explained properly.

[css-meeting-bot]

The Timed Text Working Group just discussed Make "Using computed attribute values" normative w3c/dapt#256, and agreed to the following:

• SUMMARY: @nigelmegitt to commit proposed change and @cconcolato to approve before merge

The full IRC log of that discussion

<nigel> Subtopic: Make "Using computed attribute values" normative #256
<nigel> github: https://github.com//pull/256
<cpn> Nigel: Andreas approved this PR. Cyril may be requesting changes
<cpn> ... Your comment wasn't about a change introduced in the PR, about using computed attribute values
<cpn> ... In section 6.4, there's a paragraph about computing times to div elements
<cpn> ... I tried another rewrite. What to do?
<cpn> Cyril: Your rewrite looks good
<nigel> SUMMARY: @nigelmegitt to commit proposed change and @cconcolato to approve before merge
<cpn> Nigel: Any other comments on this?
<cpn> (nothing)

[cconcolato]

Overall, I'm approving the PR to unlock the CR.

That said, I don't think it is a substantive change because the changes are about examples. The section contains a single MUST statement that is not changed.

Anyway, I think we could improve the section editorially later on. I find it scary for implementers because the boundaries of the implementation requirements are not clear. There are many examples (xml:lang, begin with nested divs, styles), but the concrete normative text is not visible enough, upfront. As I understand it, there are only 2 concrete consequences on implementers:

• use of style TTML2 properties, and there are 2 cases: either the implementation cares about them and it needs to implement proper inheritance; or it does not care about them and can fully ignore them
• use of nested divs (which I'm not a big fan of...). Implementations don't have a choice. To be compliant, they have to apply the inheritance model on intermediate divs.

[nigelmegitt]

That said, I don't think it is a substantive change because the changes are about examples. The section contains a single MUST statement that is not changed.

Thanks @cconcolato - the reason the section was made normative was because it had a MUST statement in, but was previously marked as non-normative.

there are only 2 concrete consequences on implementers

Not just style properties and nested divs, also timing properties and anything else that can be put on a div, like xml:lang.

Thanks for the approval, I will merge, and if more problems or improvements are identified they can be dealt with in new issues and pull requests.