[Prev][Next][Index][Thread]

Re: ERB call on addressing



Paul Prescod <papresco@calum.csclub.uwaterloo.ca> wrote:
> Doesn't it make sense to be able to make Panorama-like browsers that can
> intelligently navigate documents on DynaWeb-like servers without express
> communication between the vendors?

Yes.  In fact, this happens today.
It's very simple, too -- perhaps too simple.  Each chapter has an explicit
link to the next chapter.  The browser does not need to have any knnowledge
that the object being viewed is a piece of a larger structure.

This does not, of course, give a Panorama Navigator window for the
larger document.  If you want to do that, you'll need to specify an
_awful_ lot more than saying what you want the URL to look like.

> I would expect to be able to tell my
> cleint "go to the next chapter" and have the client and server agree on
> the protocol to fetch that.

This is a very, very big can of worms.  I would like us not to open it,
because I overslept and have not had lunch, and it might give me ideas.

See the expired rel-rev Internet Draft for how to do this in HTML;
it's coming back as part of Web Collections.

> I could buy the argument that we should not standardize query
> parameters in XML, maybe, but are you advocating that they should
> never be standardized?
Never is a long word.  We are not today in a position to standardise
what Panorama would need, if modified to handle XML and to have a
notion of next and previous chapter, section, paragraph, part and book.

Note that next/prev chapter is insufficient if the chapters are
themselves too large to fit in memory, e.g. for an aircraft
maintenance manual or for a typical TCIF document, or if there
are nchapters, but there is a different sort of division, as for
example in an EPSYS parts catalogue.

Consider a dictionary of biography -- you could go to the next section
in the current biography, or the next entry, or the next related
entry in a list generated by some earlier query, or the next entry for
someone who was alive at around the same time, or the person who
was born next, or...  Should we standardise a list of those?
I don't think so.

> Surely we should at least standardized a "base" simple language and
> Panorama and DynaWeb can innovate however they want around that
> standard core.

XML and URLs is a sufficient base language for the products to
work together at that level.  (Gavin, yell if I am wrong!).
Note that DyaWeb is a server and Panorama a client.  But as long
as DynaWeb serves up XML using HTTP, XML-Panorama and XML-DynaText
could interoperate without difficulty.  This is the magic of opaque
strings handled by the server.

Lee


Follow-Ups: