UI through annotations

Francois-Rene Rideau (rideau@clipper.ens.fr)
Sun, 29 Jan 95 23:50:05 MET


> On a slightly related note, I've been thinking about your UI proposal
> again, and how you should be able to ask for services without knowing all
> the particulars of how it will go about happening. My question is: why
> limit this great idea to the UI? The most important point you brought up
> (IMHO) as far as this is concerned is, "I'll say to the interface,
> 'Hello, this is me; could you please give me an object with such
> properties?' and the interface would do something, and say, 'Oh, yes,
> here it is!'" I don't see anything here limiting this to the UI, which
> should be something to keep in mind while designing the rest of the system.

Well, I completely agree; annotations could be used to customize just
anything, even the way other annotations are to be resolved, which is quite
a problem in a dynamic, distributed system !!!
For a moment, I even thought about putting the HLL stuff under the HLL
tree. But I think that would be wrong: I should rename "HLL" as a "semantics"
project, and keep the UI outa it, or just create an "annotations" subproject
under the HLL and have the UI collaborate with it. But UI has to do both with
the HL part and the LL part. UI is at the interface not only of humans, but of
a lot of parts of a computing system (though not all of them), which made
bright computer engineers to center their system exclusively on the UI (sigh).
UI has got much more than annotations; annotations is just the standard
high-level way to access it.
So yes, your remark was quite useful, because it helped me cristallize
these thoughts. Thanks !

PS:
* you can put this as a follow-up for my message in your UI page
* I haven't seen the new UI page yet. Have your thought about using
Hypermail to HTML-ify your UI mail archive ?