Re: Link-2: Pseudo-elements

At 13:15 21/05/97 -0500, David Durand wrote:

>Perhaps we could declare that pseudo elements are not counted at all: we
>would need a special keyword to address PCDATA, and could then treat all
>PCDATA in an element as a single string. This is unambigous, but also very
>counterintuitive:
>
><p>something <hp>highlighted</hp> for effect</p>
>
>would have (for addressing purposes) two addressible PCDATA chunks:
>"something  for effect" and "highlighted". They would be addressed as "The
>PCDATA in <p> and the PCDATA within <hp>, and could be further addressed
>into as via token numbers or offsets within the PCDATA.
>
>This addressing model is less obvious than the pseudo-element model, but
>(if used with some tokenization scheme) is much less sensistive to being
>put off by whitespace anomalies.

The other possibility is to count within all the data character
*descendants* of an element instead of within all the data character
children of an element. This avoids the white space anomalies but doesn't
run into the issue you raised.

James

Received on Thursday, 22 May 1997 02:38:02 UTC