Re: [xml-dev] Use DTDs!

[Gareth Oakes <goakes@gpsl.co>]

Hi Rick,

> I think there are three uses cases for named characters apart from so-called delimiter escaping:

> 

> 1. The glyph is visually ambiguous. Such as nbsp.  (The ZWNJ would be another.)

> 2. To indicate that a specialist character is involved, not the everyday one. A non trained user would not know which to pick.  The MathML set has examples. outside of maths, tne dashes are another example.

> 3. in case the font does not have the glyph, or the user cannot figure out the name. Such as Euro, or Uuml.

I think #1 is the best argument, but there is also a strong case for #2 in certain subject areas. You give two good examples and I offer up an anecdote; we had some interesting challenges with Greek letters, letter forms, and diacritics regarding whether the intent for the letter was for maths or everyday modern Greek usage.

BTW, quite commonly we come across what I would consider either a fourth case or an extension of #2 or #3. I note this only for the sake of completeness. This fourth case is where a special symbol is required that is not available in Unicode. This may be something like a hazard symbol or the Creative Commons (cc). I would consider these symbols as ideographs rather than just inline images.

Irrespective of the above, my current thought process is that XML applications should represent the characters as best suits the user and their actual use case(s). e.g. highlight "weird" characters, use an alternative glyph, warning dialogs. I guess that applies generally to any application that needs to support a wide character repertoire. Does Topologi have a neat solution? :)

// Gareth Oakes

// Chief Architect, GPSL

// www.gpsl.co