Re: REL and REV attributes (Was: More comments on HTML 3.0)

Craig Hubley (
Sat, 29 Apr 95 00:21:29 EDT

> What happened in the hypertext community "since 1967" may not apply to
> the web community. In particular, most (all) hypertext systems since
> 1967 were closed systems with a relatively small number of authors
> working together in a small community of practice. In such a

Very true, although HTML is not simply an artifact of 'the Web
community' but a general purpose hypertext authoring language that has
many applications in such 'small communities of practice' as well as
its public information distribution role.

As the Web matures I expect to see many such communities emerge, each
possibly paying little attention to each other's authoring practices
but expecting to take advantage of a common/amortized technology base.

I think a core 'link type definition' facility that provides this but
does not prejudice authoring practices is the best way to make it happen.

> community, ad hoc link-type keywords might be reasonable, since the
> community can grow to understand the meaning of tags through other
> communication paths.

These aids are available to WAN/Web users as well, in the form of
other users, online help systems, detailed descriptions of the semantics
of a link that browsers may bring up when a user lingers over it for a
while, etc.

> Personally, I'd prefer having link types registered than ad hoc, and

Registration is reasonable for those types that have to be machine-
interpreted and are likely to come into conflict with others' usage.
Like the 'well known ports' in TCP/IP, I don't think this always is
useful. Usage 'ad hoc' should be allowed to precede any registration,
if only so the registration space does not become crowded with names
that are rarely used and not well thought out in the beginning.

> to treat them as programming-language keywords rather than english
> keywords. For example, I would prefer that a French or Japanese
> browser still use 'back' for the operation rather than the French or
> Japanese translation.

Interesting point. Here you are right at the crux of the matter: are
we specifying the type for interpretation by the browser/software/etc.
or for the human ? We probably have to differentiate 'type of link'
and 'name of link' to make this happen. As in programs in other
languages, the type name is usually in English, the variable in local

For registered/known link types, type could be translated into a name.

Craig Hubley                Business that runs on knowledge
Craig Hubley & Associates   needs software that runs on the net     416-778-6136    416-778-1965 FAX
Seventy Eaton Avenue, Toronto, Ontario, Canada M4J 2Z5