Re: S.O.D.A. --> Relational GUI

From: Richard MacDonald <macdonaldrj_at_att.net>
Date: Tue, 24 Jul 2001 00:55:59 GMT
Message-ID: <zI377.16551$gj1.1710032_at_bgtnsc05-news.ops.worldnet.att.net>


"Mikito Harakiri" <mikharakiri_at_yahoo.com> wrote in message news:bdf69bdf.0107230837.28cfa686_at_posting.google.com...
> "Richard MacDonald" <macdonaldrj_at_att.net> wrote in message
 news:<7TK67.15377$gj1.1599296_at_bgtnsc05-news.ops.worldnet.att.net>...
> > >
> > > It maybe just a terminology, but let's talk about Relational Access to
 GUI
> > > engine, not relational database, because the latter is usually
 associated
 with
> > > some storage on disks. First, we must establish the schema, and this
 is
 probably
> > > the most challeging part. My naive attempts to model anything that is
 realistic
> > > from GUI perspective ended up with very unnatural schemas. Nesting GUI
> > > widgets/layouts is the biggest problem of all. So I wouldn't try to
 fool
 your by
> > > just throwing a couple of tables ala Countries, Comboboxes, etc, and
 writing SQL
> > > statement. In short, if I knew how Relational GUI engine should work,
 I
 would be
> > > already busy implementing it.
> > >
> > Ok, it just took me a while to realize you were thinking of a relational
> > schema for
> > GUI widgets and representing an application's GUI in these tables.
> >
> > I don't see why this would be so hard to do. Its pretty easy to take a
 good
> > existing
> > GUI framework and simply map its attributes via an O/R mapper. That
 would be
> > a good starting point and would satisfy your SQL query desires.
> > Sure, SQL isn't going to be able to handle the nested guis (the BOM
 problem)
> > though.
>
> Disagree. Paraphrasing: "we'll just get some [arbitrary] physical
> implementation, and put relational interface upon it [make it nice
> logical model]". Also, SQL or not, reconciling nested widget
> structures with relational model requires some nontrivial effort.

I agree with your disagreement. But you're putting the worst-case spin on it. My spin was that if we had a "good" existing GUI framework, we would have a "good" *starting point*. And the O/R mappers handle the composite pattern with ease. Received on Tue Jul 24 2001 - 02:55:59 CEST

Original text of this message