[sc34wg3] SAM issue: locator-notation-support

[Martin Bryan]

Steve

In response to my

> >HyTime locators can require more than a string: they can be defined in
terms
> >of finite coordinate spaces, property values, lists or notation data as
> >well.
>
> But surely all of these things can be represented as strings at the end
> of the day, otherwise it wouldn't be possible to interchange them in the
> form of a HyTime document (which is just a string, right?).

This depends on whether things with tags in are considered to be valid parts
of strings or not. Most of the HyTime addresses have at least one level of
embedded tagging. For things like FCS and lists you have a more complex
substructure. For property values you have QPNs (qualified property names).
Notations can suffer from the "valid character set" problem in that they
might include control codes or reserved XML symbols. (Its year's since I
went through all this stuff though!)

> The question is, is it enough to have a "HyTime" locator notation, or
> should we have several notations, one for each *kind* of HyTime locator
> (queryloc, treeloc, nameloc, etc.)?

I would prefer not to have them defined separately if we can get away
without it, even if it meant restricting the list of permitted location
types to those that can be defined using strings and stating in the text
which elements those strings have been derived from for each of the types.

> Perhaps we need the input of Eliot or Peter N on this.

Most definitely.

Martin