[REBOL] Re: About CONTINUATIONS
From: robbo1mark:aol at: 25-Feb-2002 12:50
*SPEED* is always relative and dependant upon many factors, Interpreted REBOL code is
slower than native compiled code, but I wasn't talking about speed.
I know SPEED was one of the crucial factors in the change over from REBOL 1.x to REBOL
however I was enquiring why these LANGUAGE features were dropped, which Holger answered
excellently, my original line of enquiry was wouldn't have been possible to try to improve
and optimise the continuation model citing other functional languages that appear to
have solved some of Joe Marshall cons' rather than drop it altogether.
Holger splendidly outlined the reasons why & the design trade offs involved.
Maarten Koopman also pointed out that MzScheme achieves a lot of the things which Holger
said would be much more difficult by not using the current version 2.x stack based implementation
model. MzScheme is based on the continuation model, there is also a native code compiler
MZC for MZScheme. There are always alternative avenues of possibility to be explored
/ experimented with.
For lots of reasons the current 2.x model is superior to REBOL 1.x but as Holger states
it precludes some other things.
Maybe opening up the abandoned REBOL 1.x tree would allow those interested to fully explore
the benefits and pitfalls of the continuation model? Perhaps it might be easier to develop
a REBOL compiler from the knowledge and sources available for Scheme / Lisp / ML compilers
via this route / source tree?
I just wish it wasn't so difficult to explore these issues in REBOL.
In a message dated Mon, 25 Feb 2002 11:53:54 AM Eastern Standard Time, Gabriele Santilli