Re: STk *and* TCL?

From: Erick Gallesio <>
Date: Tue, 04 Oct 1994 15:25:32 +0000

> Som people around here don't want to give up TCL, they think that the
> TCL syntax appeals to some users more than Scheme does, which I guess
> is hard to deny - though it certainly doesn't appeal to me. They want
> to be able to execute TCL code and STk code during a single run. This
> would also be useful if you want to execute existing TCL code. Any
> ideas about what it would take to do this?

I also agree that a more "classical" syntax (i.e. algol like) is certainly
more pleasant for a majority of users. However I'm not sure that using the Tcl
one would be a good choice. The only advantage I see, and it is great, to this
approach is that it would simplify the port od Tcl/Tk application to STk.
However, a perfect compatibility would be probaly difficult to achieve. In
fact there is one experiment with a language called "Rush" which tends to do
such a thing. Rush is implemented as a translator to Scheme (Scheme->C).
One thing which seems more promising is to try to have a "true" language
with true types and true control structures. This would lead the loss of Tcl
compatibility (but is it a problem if it is only partial?) but would permit
something which is more coherent. One possible way is to define a language
close from Dylan since this language could be seen as a Scheme + a CLOS like OO
system with a algol like syntax. In fact the fisrt definition of Dylan uses a
Scheme like syntax and the actual one is an algol like syntax. Marlais is
an available (partial I think) implementation of Dylan which recognizes both
syntaxes (but not the two at the same time :<).

This second way is close to the point of vue that Stallman exposes in his
famous news. I think that it could be interesting to look at it and I would
probably have a look to it in the near future (after 2.2).

I wait for the advice of the mailing list subscriber for this point.

References to the Rush papers

  - A New Architecture for the Implementation of Scripting Languages.
     to appear in USENIX Symposium on Very High Level Languages.*

  - TC: An Efficient Implementation of the Tcl Language
     UC Berkeley Technical Report #UCB/CSD-94-812*

                -- Erick
Received on Tue Oct 04 1994 - 15:25:33 CET

This archive was generated by hypermail 2.3.0 : Mon Jul 21 2014 - 19:38:59 CEST