user language revisions, was: crazy run-time fluctuations (mostly super-slow) -- a bit more data
Subject: user language revisions, was: crazy run-time fluctuations (mostly super-slow) -- a bit more data
From: Bill Wood
Date: Sun, 23 Oct 2011 16:15:39 -0500
On Sun, 2011-10-23 at 11:55 -0600, Robert Dodier wrote:
. . .
> Well, have at it. We have kicked around various ideas for improving
> the user language. My favorites at the moment are packages or
> namespaces and lexical scope. Make whatever suggestions you want.
> I am interested to hear about it.
>
> If a user language runs on top of the same Lisp core, then I would
> say that's not too disruptive.
Deciding on a nice-to-have feature or two (or 3, or 4, ...) is one
thing, but a serious language design effort is something else
altogether. What are the "right" primitives for a CAS? The "right"
combining mechanisms? What would the "best" API look like? Is the
focus to be on interactive problem exploration or on disciplined
mathematical algorithm design and implementation? Language design is
always difficult, always filled with compromise. The best examples
exhibit design integrity, and are rarely if ever done by committee.
That said, it is true that lisp's "ball-of-mud" character has stood the
tests of time and fad well.
--
Bill Wood