Re: [Usability] making the conceptual model more concrete

[Gregory Merchan <merchan phys lsu edu>]

On Sun, Jun 29, 2003 at 11:15:36AM -0400, Havoc Pennington wrote:
> On Sun, Jun 29, 2003 at 12:30:04AM -0500, Gregory Merchan wrote: 
> > The last time I tried this [1], I got monkeycrap [2] flung in my
> > face.
> 
> Well, whatever. People flame, it's the internet.

And I'm not flaming you for bringing this up now. See my response to Jeff.
(Perhaps I should have added, "Let's proceed without a repeat of that.")


> Anyway, the proposed conceptual model change is a rationale for some
> UI reworking. What are the concrete changes that are justified by said
> rationale? This is a fairly focused question it seems like.

Calum calls it an example, and I think he means that. The possible model
depicted there[1] is incomplete.

More thorough and detailed models are provided in IBM's CUA[2]. In it there
are at least three models: user's, programmer's, and designer's. The
summary of the designer's model begins here:

  http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/F29AL000/1.5?SHELF=CEESL002&DT=19921204095534


> I'm not asking for a mailing list thread on it, simply suggesting that
> a good descriptive writeup with pretty pictures might make it a lot
> easier for developers to visualize. Right now I'm not sure we
> understand exactly what those model diagrams imply.

It's as difficult a task as writing a specification for anything else.
As with programming, it often seems easier to just work on the parts with
just a notion of the whole than to specify the whole then work on the parts.

I don't know about anyone else, but for most of the things I've brought up,
I've had a notion of the whole behind them. Because of different notions of
the whole, we have flamewars about the parts.

It is vitally important to have 1) a consistent model 2) that can be
implemented now, and 3) cab carry GNOME into the near future without overhaul.

So, yes, we need a documented model and then specific bug reports etc.
to move towards it. Right now a few of us have models in mind that we know
overlap to some extent. Eazel seemed to have one, but never (AFAIK) made it
public. Ximian and RedHat, each, probably have some sort of implicit model.


Cheers,
Greg


[1] http://developer.gnome.org/projects/gup/articles/guadec-2003/guadec-paper_html_m448a52dd.png
[2] http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/F29AL000/CCONTENTS?SHELF=CEESL002&DN=SC34-4399-00&DT=19921204095534