Re: IDs - make them case sensitive

Rick Jelliffe <ricko@allette.com.au> wrote:

> > From: Murray Altheim <altheim@mehitabel.Eng.Sun.COM>
> 
> > The idea of not being
> > able to have what is essentially a numeric type (ID) not be able to
> > contain a completely numeric value seems somewhat strange.
> > Like my Sun ID# is all numbers, my SSN# is all numbers and dashes, etc.
> 
> My guesses are it was
> 
> * to stop  "if a<3 then blort" causing start tag recognition, because
> of the extreme conservatism of the original design (all those context
> checks on delimiters),
> 
> * to avoid names being numbers, as a matter of human-readability.


My guess is that: 1) digits cannot be name start characters
so that numeric character references (&#123;) can be 
distinguished from named character references (&#ABC;); and
2) element names (id values) are NAMEs because every other
kind of name is a NAME (element type names, entity names,
document type names, attribute names, etc.).


> If people really want IDs to be able to start with digits, they can
> also demand of their local WG8 delegates    
>    "We want IDs to be 'name tokens' not 'names' !" 
> or
>    "We want a new type of ID (NID?) that uses 'name tokens' not 'names'!"
> or even
>    "We want a new type of ID (el CID?) that uses CDATA rather than 'names' !"
> or
>    "We want lexical typing, normalisation and uniqueness-checking built
>    at language level into SGML !"


--joe@art.com

Received on Tuesday, 1 July 1997 18:08:54 UTC