Re: [i] XPointer and addressing in OHP-Nav

From: Lloyd Rutledge (Lloyd.Rutledge_at_cwi.nl)
Date: Tue 21 Jul 1998 - 08:32:46 CDT


On Tue, Jul 21 1998 Peter Nuernberg wrote:
> However, we should be careful about statements like this:
>
> > I think the interoperability with the Web is the key issue here. We
> > should not worry about the fact that XPointer is only applicable to
> > XML and HTML...
>
> I would say that _openness_ is the key issue here, since we are the
> open hypermedia systems group, and not the WWW improvement community
> or W3C standards implementers association or some other such thing.
> It is in principle bad to use non-open standards in an open system.
> So, maybe we can support XPointer as a type of LocSpec, or maybe
> better we should use LocSpecs between the server and the wrapper and
> just build smart wrappers for WWW tools that translate between open
> LocSpecs and closed XPointer specs.

Sure, "openness" is the key issue in general for this group, but interoperability with the Web is an _unavoidable_ issue -- perhaps _the_ unavoidable issue. And sure, we should avoid statements that may imply we wish to implement the Web simply for the Web's sake. But we should also avoid thinking that the Web is bad and we can't use it and that we are making something better than the Web that will exist without it. For better or for worse, as of now nothing in hypermedia that doesn't happen on and with the Web will ever matter anymore. Not that the Web is perfect, but it is the medium on which everything will happen, or at least on which everyone will expect everything to happen, which is the same thing. The Web, like air, is omnipresent, and living within it is the difference between breathing and not.

The bright side is that the Web is adapting and is adaptable and we can influence it, which is what our involvement in XLink and XPointer is all about. If there is something about XLink or XPointer that can't apply to us, then this is something that the folks in the XML working group want to know (which should soon have an XLink/XPointer offshoot WG), and now is exactly the right time to tell them.

The Web may not really be "open" (now), but the folks in the XML WG would emphatically deny wanting it to be otherwise and will listen (probably/hopefully) to what we have to say about the issue. I'm not really sure what the statement about "non-open standards" meant, since, although HTML may be monolithic and unforesightfully restrictive, the word "open" has always been the battlecry of SGML, and thus of its progeny XML, which is the current focus of developing the Web within W3C (and also within that other unignorable 400 pound gorilla, Microsoft).

With what we've been doing for the past four years and with the goals of XML in general and XLink and XPointer in particular, we are in a very good position to influence the Web to do many things we'd like it to do. Our being effective as such involves holding a balance that touches neither extreme of being either gratuitiously Web implementing or non-(widespread)implementably purist.

But having just been indulgently wordy addressing spiritual issues, I'll try to throw a bit of substance at the end of this posting. First of all, I agree with Jacco's rewording of the conclusion,

> OHP-Nav could use XPointer for linking into XML/SGML/HTML documents,
> especially in applications for which interoperability with
> WWW-applications is an important issue. It could use any another
> mechanism in all other cases.

but would add that the XPointer group may be influenced to add constructs for these other mechanisms. Such work would be helpful, for example, in including portions of media objects, such as video clips from large video files, into Web presentation. XPointer is based (loosely) on HyTime, and its head Steven DeRose literally wrote the book on HyTime. HyTime has constructs for pointing to non-SGML data. Perhaps with our input such constructs and others could be put into XPointer, and XPointer could be made to better fit the concepts of good open locating we've been working on.

Sigi (or whomever), could you make a list of specific requirements that XPointer would need to fulfill to be able to solve all the needs that have been established with your experience in developing OHP and OHP-Nav, and then we could send this on to the XPointer people (or for starters, to Steven DeRose) and see what their response is?

-Lloyd



This archive was generated by hypermail 2.1.5 : Tue 13 Aug 2002 - 07:21:02 CDT