Re: [xml-dev] SGML default attributes.

[Peter Flynn <peter@silmaril.ie>]

On 05/04/2016 11:23 PM, Eliot Kimber wrote:
> If you look at an environment like PubMed, where you have many 
> loosely-governed and coordinated publishers contributing documents
> that are all supposed to be in the same document type you start to
> see the problem. They are all (or mostly all) DTD valid. Many claim
> to be valid to the JATS or NLM DTD, but they have modified these DTDs
> locally.

There seem to be two reasons for people doing this (and I'm not
referring specifically to JATS or PubMed here):

1. Ignorance (and I don't mean that pejoratively, I just mean lack of
knowledge). More and better author education, and definitely more and
better authoring tools are needed, especially editors.

2. The base document types (if I may call it that; the intended,
unmodified DTDs) did not provide markup for something the author
believed warranted its own identity. Without a full analysis, it's not
possible to say if these are genuine defects in DTDs, or simply wishful
thinking by the authors.

This is not to excuse the interference with a DTD, but having been
called in to fix this kind of thing on many occasions, I'm sorry to say
that the involvement of authors in document type creation does seem in
some cases to be a low priority.

> Many have modified the DTDs and given them new public IDs
> (as they should) but then there's no way to know what they really
> are.

Reintroduce FPIs and require registration, with mandatory resolution via
the registries to a human-readable resource (via URI) that provides
documentation, but without the defererencing being required for
well-formedness or validity. Not a hope.

> It's a mess that the engineering team at PubMed went to heroic
> efforts to try to manage through transforms and so forth.

Instead of outright rejection. Suppose someone submitted a paper in
corrupted OOXML that Word could not read?

///Peter