The first token in a processing instruction should identify what the
"target" of the processing instruction is. (This might be identified
formally in a notation declaration too.) So you can have
a million different types of processing instructions added, one
for every flavour of Java or Javascript or ECMAscript or everything
called dynamic HTML or serverside includes. All at the same
time, all clearly labelled.
e.g. {sun-java1 ...}{ms-java ...}{sun-java1.1 ...}
And the processing instructions allow markup that does not follow the
bondaries of the elements. You can start something in the middle of
one string in one element and end it somewhere in a cousin.
There is a place for putting scripts in nice elements, for those
that act globally, or get invoked by user agents, but inline
scripts that merely process some text automatically are better done
in processing instructions rather than corrupt the element structure.
Rick Jelliffe