Re: Dreaming About Redesigning SQL

"Mikito Harakiri" <mikharakiri_at_iahu.com> wrote in message news:sUbnb.19$nM4.107_at_news.oracle.com...
> "Bob Badour" <bbadour_at_golden.net> wrote in message
> news:3rudnXhh4tAyFwGiXTWJkQ_at_golden.net...
> >
> > > > One can describe anything relationally. What makes you think it
> > impossible
> > > > to describe a GUI relationally?
> > >
> > > I must correct my statement, of course: So far nobody succeeded doing
> > > so.
> > >
> > > Application developers routinely describe GUI like this:
> > >
> > > <table>
> > > <tr>
> > > <td>
> > > User:
> > > </td>
> > > <td>
> > > <input type="text" name="usr" value="<%=usr%>">
> > > </td>
> > > </tr>
> > > <tr>
> > > <td>
> > > Password:
> > > </td>
> > > <td>
> > > <input type="password" name="pwd" value="<%=pwd%>">
> > > </td>
> > > </tr>
> > > </table>
> > >
> > > which is just 4 GUI elements in 4 cells layout table, and this
> > > approach proved to be quite sucessfull. For comparison, try describung
> > > this trivial GUI page in relational to see what I mean.
> >
> > What's difficult about that? It's just two tuples.
>
> Even a sinlgle tuple. On the data side. On the GUI side it is not. You
have
> to specify 2-dimensional spatial relations among the GUI elements, and
this
> is where the problem becomes serious.
>
Why? The spatial relationships are just data that are fed to a layout engine, and can be represented as tuples: e.g. absolute x, y positions, position relative to another widget, or geometry specifications (snap to edge, etc). Absence of layout is just a directive to use defaults, as in the example above. Received on Mon Oct 27 2003 - 23:48:29 CET