The Dave Scripting Challenge

Dave Winer dave@userland.com
Sun, 26 Aug 2001 14:31:25 -0700


Yes, now I think you get it.

The language syntax, awkward as it is (to the point of being unusuable
imho), is a perfect fit for the limited verb set that you're exploring. That
they have an open architecture for verbs is not a big point. If they had
tightened the noose even further you'd get REST. Nice in theory, bizarre in
practice.

Dave


----- Original Message -----
From: "Jeff Bone" <jbone@jump.net>
To: "Dave Winer" <dave@userland.com>
Cc: <FoRK@xent.com>
Sent: Sunday, August 26, 2001 2:25 PM
Subject: Re: The Dave Scripting Challenge


>
>
> Dave Winer wrote:
>
> > With all due respect, you totally missed the point of AppleScript.
> >
> > Look at the object model for controlling external apps.
>
> I think I got your point --- I just don't think the mapping to REST is as
> direct as you think it is.  I think this is fundamental to the differences
> we're talking about, so let's press on...
>
> Sanity-check me here:  the Applescript object model exposes application
objects
> inside applications and their documents to scripting via dictionaries.
These
> dictionaries define the classes of the objects exported, their properties
and
> the commands which operate on them.  Each application has its own object
model
> (set of nouns) and command set (set of verbs) though there is significant
> consistency between apps with respect to the latter.  All objects live in
a
> hierarchical relationship;  "[e]verything belongs to, is related to, is
> contained in, or is part of something else."  The coordination paradigm is
to
> have scripts navigate this "chain of command" ultimately invoking
operations on
> some object of interest.
>
> Similarities:  I'll admit that the "chain of command" container / part /
> relation hierarchy has some absolutely striking similarity (it's
identical!) to
> the kind of resource modeling paradigm I've described elsewhere.
>
> Differences:  even though there is a significant amount of consistency in
a
> basic set of commands most applications support, it differs in that it's a
> larger set of verbs, not all applicaitons support all (or any) of them
> particularly, and each application can actually define its own syntax for
a
> given command.  Not only are command implementations free to decide their
own
> arity and signature, they're even able to *change the syntax of the
language*
> on a local scale.
>
> That's a HUGE difference.  I'm not saying it's bad or wrong or ugly, but
it's a
> fundamentally different approach to coordination and reuse than a system
> providing a consistent compositional algebra for wiring together objects
with
> identical generic interfaces.
>
> You do get what I'm saying, don't you Dave?  I'll keep poking around in
> Applescript some looking for more Zen.  :-)  I've no doubt there are even
more
> interesting things lurking around in there...
>
> jb
>
>
>
>
> http://xent.com/mailman/listinfo/fork