On Mon, Feb 23, 2015 at 7:39 AM, James Cummings
- don't overuse @xml:id. Only use @xml:id for referencing fragments of a document. The value of @xml:id shall not be parsed into components during document processing
I know I'm probably wrong but because I'm lazy I tend to prefer over-use of @xml:id. I understand it might not be theoretically sound, but then again I'd also like human-parseable values as well. So I might have IDs like poem1, poem1-stanza1, poem1-stanza1-line4, merely to facilitate other people pointing into my document and talking about it (or grabbing bits). Ok, I understand they can do so through other methods but it is a lot easier to say give me foo.xml#poem1-stanza1-line4 than to do a proper XPointer to it.
Similarly, I've argued elsewhere that facilitation of stand-off digital editions is much more pragmatic if the underlying edition has provided IDs on the smallest reasonable level of granularity. I have put IDs on every single word before precisely for this purpose because if words are interspersed later then the IDs have not changed. If a name goes from #w10 to #w13 and you realise you left out one of his middle names, then adding in #w12a is mostly side-effect free.
I don't think these would qualify as over-use in Stefan's formulation,
since they're providing fragment identifiers. If I understand right, he's saying "don't overload the semantics of xml:id", i.e. don't use them in a such a way that you expect to be able to extract information from them; only use them to identify bits of your document.