Re: [dita] LwDITA markup inconsistency, missing keyref on highlight elements

[Eliot Kimber <ekimber@contrext.com>]

I would definitely vote to allow keyref: more consistency and less surprise. 

It's a pretty easy constraint to impose if anyone cares that much.

Cheers,

E.

--
Eliot Kimber
http://contrext.com
 

ïOn 5/24/21, 1:57 PM, "Robert Anderson" <dita@lists.oasis-open.org on behalf of robert.dan.anderson@oracle.com> wrote:


    Hi all,


    Perhaps I should stop digging through DTDs and RNGs, but while doing so I've found another inconsistency. This one is actually an issue for LwDITA rather than for the base spec, but addressing it likely impacts the base spec.


    In LwDITA, all phrase-level reuse is done with the keyref attribute, rather than conref, as part of an effort to make sure there is only one way to reuse any type of phrase. This means that the LwDITA modules expect keyref to be on highlight elements such as
     <b>, <i>, and <u>, as well as the upcoming elements <em> and <strong>. The %variable-content; entity in here adds keyref to five highlight elements:
    https://github.com/oasis-tcs/dita-lwdita/blob/master/org.oasis-open.xdita/dtd/lw-highlightDomain.mod#L45



    The problem is that in the base DITA vocabulary, none of these elements allow keyref, so specifying it on LwDITA highlight phrases means that those elements are no longer a valid subset of DITA. It's not there in

    DITA 1.3 grammar <https://github.com/oasis-tcs/dita/blob/master/doctypes/rng/base/rng/highlightDomain.rng#L105> and it's still not there in the 
    DITA 2.0 definition <https://github.com/oasis-tcs/dita/blob/DITA-2.0/doctypes/rng/base/highlightDomain.rng#L114>.


    Digging deep into memory, the only reason I can think of that it was left off was to nudge authors into using more semantic elements for reuse. These elements all specialize from <ph> which does allow key references, so it would be legal to add it. Given the
     LwDITA use case, it seems like at this point we should probably let these elements use keyref, unless there is a compelling reason to keep it off. If we do decide to keep it off, that means a fair bit of churn in how reuse is managed in the context of LwDITA.


    Thanks,
    Robert