2. We might call these "selection" anchors.
3. The syntax is selection anchors cannot include line numbers of rendered
document as this is not constant for all viewers. They should identify
both where to start and where to end.
4. Can we base the selection anchors on SGML Object Identifiers?
Are there other standard conventions that may apply?
A syntax like ##[[count][markup]]*,[[count][markup]]* might work, e.g.
##3<H1>2<H2>3<p>,2<p> references the 3rd to 5th paragraph in the
second subsection of the third major section (H1). A count without
a markup might represent a word/element or a character count (both
might be useful). A start pattern and end pattern may also be useful.
I would thing that reasonable conventios for addressing SGML segments
would exist. The SGML Object Identifier for example, is intended
to do this-- but I've misplaced the spec and cannot say how well it
succedes.
5. It would be useful if browsers could generate selection anchors
for HTML segments selected (highlighed) in the browser. For example,
in collaborative hypertext, the user could select the phrase where they
want to add a link. On a HTTP GET, the selection anchor on the
HTTP_REFERER would indicate what HTML segment was highlighed by the user.
6. If there are browser or server implimentor looking to add "selection"
anchors we should define the conventions. Most current browsers, as
far as I know, do not send the #anchorName in the HTTP GET or POST
and will not, therefor, work as is with selection anchors.