Re: @xml:base with @rendition (and maybe other pointers)

> On May 3, 2017, at 6:26 AM, John P. McCaskey =
<[email protected]> wrote:
>=20
> Another reason to design for internal renditions only:
>=20
> .,..
> Corpus-wide formatting changes should be done with stylesheets, not =
external rendition pointers.

While I sympathize with this point of view, up to a point, a few =
technical points
which point the other way should be borne in mind:

1 One of the most fundamental decisions in the design of URIs is that =
they
make no distinction in principle between pointers within a document and =
pointers
between documents.  (This is easily seen if one compares URIs to the =
ID/IDREF
mechanism used in SGML and XML for in-document references, which leaves
inter-document references in need of some other language for hyperlink =
addresses.) =20

2 In the design of any system, orthogonality and compositionality are =
almost always=20
betters idea than non-orthogonality and non-compositionality.   Where a =
language
accepts an arithmetic expression, it should normally accept any =
arithmetic expression;
where a language accepts a URI, it should normally accept any URI.

If one has made the decision to use URIs and not some other addressing =
mechanism,
it=E2=80=99s almost certainly a questionable idea to start imposing =
additional rules that go
against the grain of the design.  (=E2=80=9COnly same-documentl =
references here=E2=80=9D is an
excellent example of what things look like at the top of the slippery =
slope; at the
bottom are languages with lots of special cases and exceptions.  Been =
there,
done that; I assure you it=E2=80=99s not a good place to be.)

3 =E2=80=9CExternal=E2=80=9D is a slightly tricky concept for any system =
which, like XML and SGML,
allows different parts of a single =E2=80=9Cdocument=E2=80=9D to be =
stored in what the external=20
environment regards and manages as distinct entities (e.g. files).  If I =
wanted to
have my renditions in a distinct file which is referenced either as an =
external=20
entity or via XInclude, could you devise rules that could identify that =
case and
rule it out?  (I am pretty sure I can=E2=80=99t.)

The independence of element structure and entity structure is one of the=20=

fundamental design decisions of SGML, retained in XML with some =
restrictions
aimed at simplifying partial processing.   Decisions at the TEI level =
will normally=20
work best if they go with rather than against the grain of the =
technology.


********************************************
C. M. Sperberg-McQueen
Black Mesa Technologies LLC
[email protected]
http://www.blackmesatech.com
********************************************