Language implementation in Squeak (was Re: Python for Squeak)

Bijan Parsia bparsia at email.unc.edu
Wed Nov 17 01:52:02 UTC 1999


At 10:20 AM -0500 11/16/99, Jerry Jackson wrote:

[snip]
>
>I'm also very interested in this.  I spent several years long ago
>working with Xerox lisp machines (Dandelion, Dandetiger, Daybreak)
>and was amazed at how similar the Squeak environment seemed when
>I first looked at it.  I made a stab at building a Lisp machine
>on top of Squeak but decided I would wait for full closures before
>continuing.

Well, don't we have full closures, in a way? They're called methods. What
we don't have is full *block* closures. But I can imagine a Lisp (say
scheme) that compiled defines to methods. If you wanted to "save" an
"image" you could generate a subclass of the Enviroment class. or something
like that :)

>  I've given some thought as to how Smalltalk-style keyword
>messages map onto lisp and I think it can be made pretty clean.

Well *that's* certainly true. But do you mean to do source transform? I
think it's better to compile directly to the parse tree.

>I have one question for the long-time Smalltalkers...  Why did you
>move from a read/eval/print loop to the workspace model?  I
>don't have anything against the mouse but I find a listener with
>a good history mechanism is a faster interaction mode for me.

I'm not a long-time Smalltalker, but I much prefer the workspace model (as
MacGambit uses!) But, hey, so long as I'm not shackled (chancan-led?) by
ton gout, that's fine with me :)

Cheers and please do kill me for puns banned by the Academe Francaise,
Bijan Parsia.





More information about the Squeak-dev mailing list