Re: [wmii] Re: quick thanks... for wmii

On Mon, Jan 22, 2007 at 03:29:38PM +0100, Denis Grelich wrote:
> On Mon, 22 Jan 2007 14:58:40 +0100
> "Anselm R. Garbe" <arg_AT_suckless.org> wrote:
>
> > On Fri, Jan 19, 2007 at 06:40:59PM +0100, Uriel wrote:
> > > The beauty of ed is that its design is so minimalistic and its
> > > interface so elegant, that it can adapt to a completely new
> > > environment without the need of any changes.
> > >
> > > The beauty of rio terminals is that they transparently provide a
> > > plethora of new functionality for existing applications without the
> > > need to modify them or compromise their clean and simple design.
> > >
> > > Both this things are direct consequence of the Unix/Plan 9
> > > philosophy of "do one thing and do it well" and of an ecosystem of
> > > small tools that play nicely together, where each new tool augments
> > > the functionality of the already existing tools instead of having
> > > ot hack every new piece of functionality into every single tool.
> > >
> > > Vi is the total opposites, its interfaces is so ugly and awkward
> > > that the only way it can adapt to a new environment is with some
> > > serious reworking. The only way for it to not depend on a terminal
> > > model that is over thirty years old is for it to internalize the
> > > functionality of the terminal(cursor addressing etc).
> >
> > Actually I think your last statement applies much more to the
> > rio terminal than to a vt. Except the fact, that the rio
> > terminal allows the cursor being a text selection, it provides
> > no higher level grid- or cell-adressing like a vt, because it is
> > 1-dimensional and behaves much more like a typewriter/teletype than a
> > vt in my eyes. The rio terminal fails miserably whenever you run
> > a 'higher level' vt application which makes use of 2d-addressable
> > character/text output like vi. Hence it seems to me to depend on
> > technology which is already over 100 years old.
> >
> > However, because of its 1-dimensional simplicity, even hello world
> > programs in 100 years will adapt perfectly to such a terminal.
> > (In general, any program which adapts perfectly to an
> > 1-dimensional terminal, also adapts perfectly to a n-dimensional
> > terminal with n > 1).
> >
> > So the quintessence is, if you write programs which rely on 100
> > year old technology, it is quite safe that they will run for 200
> > years. It has to be proven, that vt's will last the next 70
> > years before we can decide them to be no mayfly.
>
> The problem is that having a 2D-addressable text area is not what a
> terminal should be for. Terminals under Linux are abused as some sort
> of primitive means for building a GUI from characters, which is, in
> fact, totally backwards if you can have "real" GUIs where you can draw
> pixels, lines and whatever your graphics adapter can do.
> What vt's do is forcing something modern as GUIs onto a 100 year old
> technology. A terminal should only be what the name says, the
> termination point for STDIN and STDOUT.

True, but the more dimensions are used, the higher is the
probability that it is a mayfly. If you consider how heavily GUI
design has changed within the last decade, and how heavily those
vt applications have not changed during the last two decades,
it's questionable if it already makes sense to do GUI
programming (if this should last in the long term ;))

Regards,

-- 
 Anselm R. Garbe >< http://suckless.org/~arg/ >< GPG key: 0D73F361