Interesting. A few questions/comments.

From what I can see, you will not be able to dynamically change the
page's content after it has finished loading, due to the same-origin
restriction. Is this correct? In other words, you'd have to pre-plan
for any user interactivity and send the necessary data along.

We're pursuing a similar approach with the most recent version of
MAJAX. Instead of using <div>, we're using COinS-style <span> tags
currently. (*) We tried to keep the JavaScript to a minimum - the user
shouldn't include anything more than the <script> to pull the
processing JavaScript. Then only plain HTML would be used. (By
contrast, your approach relies on hardwiring knowledge of the page
layout into your Javascript, such as "table[3]" to find out where to
insert your information - and at librarything.com at that!)

My comment, I suppose, is that it may be nice to agree on a set and
syntax of features that people could use to include into their web
offerings - not only the catalog, but other library and even
non-library offerings as well. For instance, a new book list or a list
of resources prepared by a professor for a particular class are
example applications we have prepared for Majax, and they could
certainly be implemented for/with LTFL as well.

 - Godmar

(*) See http://libx.org/majax/majaxtalk.html - BTW, Annette will talk
about this in San Jose Wednesday afternoon at the IUG meeting.


On 5/14/07, Tim Spalding <[log in to unmask]> wrote:
>
> I expect many of you will object that both are kludges. We need
> genuinely open and extensible systems.
>
> Of course, I agree with you. All the LTFL widgets are also available
> as XML feeds. And I can't wait to plug them into Evergreen, or
> whatever. But, for now, and for most people, the exsiting OPACs exist.
> There's a lot we can do with them, even so.
>
> I hope that LTFL can inspire a new generation of light-weight OPAC
> enhancements. If anyone wants to talk architecture, I'm more than
> happy.
>
> Tim
>