Scheduling subsystems (crontab, at) and the desktop

[Maciej Katafiasz]

W liście z pią, 23-07-2004, godz. 13:00, Rodrigo Moya pisze: 
[snip]
> this is great! We miss though, if you are going to use d-bus, a good
> query mechanism, so that we can query all the installed actions without
> having to start individual applications.

gah! It's there in document I linked to, I forgot to list it here. Of
course it's needed, we will most probably model it after OAF (since it's
rather easily comprehensible mechanism) + backend for storing the info.

[snip]
> as you said in the document you posted before, the UI can be dinamically
> generated from the description of the methods returned by each
> individual application. That should work pretty well for simple types,
> but not so well for other types. So I would suggest there is additional
> information on each argument, like the allowed range, the set of valid
> values, if it's a file, what kind of file it is (if it's an image, for
> instance, show a preview, etc), ...
> 
> That would solve the UI problem, since building the UI automatically for
> any desktop/toolkit would be quite easy.

Yes, but in course of brainstorming, we had been pointed out that it's
in general impossible to create working UI for any data simply from
types description (specifically, it's not about laying it out in a
pretty way, but about making sure that logic behind it works correctly,
so that checkboxes toggle sensitivity of related widgets, dates enetered
are sensible, etc.). This could be probably solved by using programming
language like python embedded in description, or maybe forging our own
declarative pseudo-language based on XML, but it's a project on its own,
and we take into account (however unpleasant) possibility of failing
here. We're gonna try it before, however ;)

> all this sounds really good, and something I have been wanting for long
> (the scriptability stuff specially), so, when are you releasing it? :)
> No, seriously, is there any code written yet? where is it?

Sorry, nothing yet. We discussed it much, but coding is only starting
now (it was stalled partly due to end of academic year)

Cheers,
Maciej

-- 
"Tautologizm to coś tautologicznego"
   Maciej Katafiasz <mnews2 at wp.pl>
       http://mathrick.blog.pl