Anchors/Fragments ?

Liam Relihan (
Sun, 23 Jan 1994 16:54:20 +0000 (WET)

hello all...

I'm doing some work in formalizing W3 by looking at it from an IR point of
view. I reckon that the objects types in W3 are those types that are
addressable by URLs:

1. nodes (which include ordinary nodes and index nodes) which may be
addressed with a path

2. virtual nodes which may be addressed with a "?"

3. fragments of nodes which may be addressed with a "#"

Now, I am unsure about the status of the "fragment-id" (or "#"). TBL's URL
spec states...

"This represents a part of, fragment of, or a sub-function within, an
object or object. Its syntax and semantics are defined by the application
responsible for the object, or the specification of the content type of
the object."

Now, if I have a URL such as the following

, from an IR point of view, retrieval of the (sub-)object identified by that
URL means the following:

1. retrieve the document identified by "" from
the server.
2. Search the document until the anchor named "parta" is found
3. Display the document and move the pointer to "parta"

My questions are...
is this sequence valid only for the HTTP scheme ?
if a fragment-id were used for another scheme (say NNTP), might the
retrieval sequence be different ?

Is fragment-id always meant for the identification of parts of objects
(sub-objects) within the context of a whole enclosing object ?

If not, why isn't "?" good enough for acquiring addressable sub-objects ?

Apologies if my questions are slightly unclear.
Unfortunately this stuff can be subtle at times :-)


Liam Relihan, | Voice: +353-61-333644 ext.5015
CSIS, Schumann Building, |/| Fax: +353-61-330876
University Of Limerick, | E-mail: