Date: Thu, 21 Jul 1994 18:42:50 -700 (PST)
From: Mitra <mitra@pandora.sf.ca.us>
Subject: Fragment ID;s
To: uri@bunyip.com
Message-Id: <Pine.3.89.9407211808.A10354-0100000@pandora.sf.ca.us>
Larry suggested putting "#" back into URLs as a fragment ID.
This was brought up and shot down at one of the WG meetings (by my
possibly flawed memory).
There are two problems with putting it in.
1) "#" is not a universal fragment id, its specific to
URLs that lead to the return of a HTML document, it has no meaning for a
binary file or even a plain text file. Note - dont confuse a URL to a
document with a content type of HTML, with a URL in the scheme HTTP,
these are orthoganal concepts, despite being frequently used
interchangably by some of those who should know better :-)
2) There were strong objections to putting fragment id's inside URLs, as
their were to putting other information in them - personally I disagree
with that conclusion, but can live with it.
3) If we want a fragment id, then lets have a decent one, I made one such
proposal at the Washington IETF - based on some of the discussion in
early Z39.50 meetings, but it was basically bumped off of the discussion
when it was decided not to put anything else inside URLs. A decent
fragment id, needs to be able to express such things as bytes 1000-2000
or the last page, or the last frame, etc. IMHO the WWW implementation of
framgents is a bit of really poor non-generic design that certainly shouldnt
get grandfathered as standard, even if we make attempts at accomodating its
usage.
- Mitra