Restriction on PI information

[Eric Baatz - Sun Microsystems Labs BOS]

My application of XML is to markup text that is to be spoken by
speech synthesizers.  To my naive mind (I'm very new to SGML
and XML), a PI seems to be the right construct for passing
native information to a speech synthesizers, that is, instructions
in their proprietary, already existing, command set.  As I don't
have any control over the syntax of the commands I want to
pass through, I want a PI to allow the widest latitude in
the information it can handle.  The syntax in the draft doesn't
seem to allow that.

What is the rationale for the data that a PI allows?
What mechanisms can be used to make that data as arbitrary
as possible without changing the draft?

My take on the PI syntax is that the data needs to avoid
looking like the end of a PI.  Two different ways of ending
a PI (somewhat like the use of double or single quotes for
quoted data) would allow a way of getting unpalatable data
through (my program would have to generate the appropriate
one depending on what my data looked like).  Allowing a CDATA
section, would also seem to allow quoting of otherwise
unpalatable data.

Clearly, any changes from the draft would complicate the parsing.

Eric Baatz
Sun Microsystems Laboratories
2 Elizabeth Drive, MS UCHL03-207                 (508) 442-0257
Chelmsford, MA 01824                        fax: (508) 250-5067
USA                                    Internet: eric.baatz at east.sun.com


xml-dev: A list for W3C XML Developers
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To unsubscribe, send to majordomo at ic.ac.uk the following message;
unsubscribe xml-dev
List coordinator, Henry Rzepa (rzepa at ic.ac.uk)