Assembly: carry an @xml:id from audio|video|interactive[not(static)] …

[Alex-Jordan]

…over to its static sidebyside

This is discussed in #2618.

Among the PRs I expect to open and update this week, first to evaluate can be this one or #2739, in either order.

[rbeezer]

If the source interactive does not have an @xml:id this is going to make an attribute on the sidebyside that has an empty value. Then, that will wreak havoc with the various auto-generated id's.

This is confirmed by testing before/after diffs with the assembly-static format, as should be part of a change like this.

Please make the "new" attribute conditional on having an "old" attribute. And a code comment indicating necessity would be useful, so nobody messes with this unawares. Thanks.

[Alex-Jordan]

I wondered about this when making the change, but something made me think it would not be an issue. I was searching for how this was done elsewhere, and at line 879, the same thing is done for a biblio. It made me think this must be OK to end up with @xml:id="".

I'll make the change here. Mentioning the biblio instance in case you feel that is worth changing too.

[rbeezer]

Thanks for the pointer to biblio. A biblio is largely useless without an @xml:id since you should reference it via an @ref.. But I guess people will make biblio in a references just for show.

[rbeezer]

Maybe try

<xsl:copy-of select="@xml:id"/>

which might do the desired thing.

[Alex-Jordan]

OK, this has your suggestion to use copy-of. I did not know about assembly-static, good to know. I used it now, and it seems the before/after diff hits about 80 instances of a change using the sample article. The changes are hard to read, because once the parent sidebyside has an author-generated xml:id, then its unique-id changes. So then all of the descendant unique-ids change, and that clutters the diff. But aside from that, they seem to hit the right things. There are no instances of xml:id="".

The before sidebyside has only a unique-id. The after sidebyside has a (different) unique-id, a (new, expected) xml:id, and also a label. Should I be expecting the label to pop up out of this too?

[rbeezer]

There are no instances of xml:id="".

I think that is the important thing. All the clutter is to be expected.

Should I be expecting the label to pop up out of this too?

Couldn't hurt. And will mask an @xml:id if an author doesn't want them to bleed into output. I'll see about adding it as part of a review.

[rbeezer]

I added in the copy of @label, plus added the requested code comment. Thanks for firming this up.

[rbeezer]

And I fixed up the biblio version at 2f2c5c2. Thanks for pointing that one out.