Re: Client-side highlighting; tag proposal

lee@sq.com
Tue, 14 Mar 1995 15:58:36 -0500


Date: Tue, 14 Mar 1995 15:58:36 -0500
From: lee@sq.com
Message-Id: <9503142058.AA04937@sqrex.sq.com>
To: www-html@www10.w3.org
Subject: Re: Client-side highlighting; tag proposal
Cc: sjd@ebt.com

I agree with Steve DeRose that one is indeed likely to see large chunks of
text, possibly with highlighting.

Further, both EBT's DynaText and SoftQuad's Explorer and Panorama products
can do highlighting across element boundaries.

BUT: I would not suggest using HyTime TREELOC notation in a web browser,
because this would require that all the browsers treat OMITTAG in the
same way.  Currently, some browsers infer a <P> where others don't, I think,
and I don't consider it practical to try and deal with this, unfortunately.

I would like to suggest instead a hybrid approach.

[1] Allow an ID attribute on any container.
[2] Use a HyTime ladder to get to the containing element by ID where
    possible
[3] Use byte and/or (natural language) word counts within that container
[4] for multiple containers, e.g. the midst of one P unto the next, make
    multiple [3]'s within the [2].

This is what SoftQuad Panorama PRO and SoftQuad Explorer do.
The HyTime can go in a separate file.

Lee