Hi Brian,

[snip]

> I like the idea of a headless squeak that can even be used for
> scriptable shell, and more importantly for serving pages.

Well, there's headless as server (using comache or PWS), headless as cgi
runner (a la perl or python), or headless persistently running (a la
VisualWave). ;)
 
> However, I don't like the idea of differing so much from smalltalk
> syntax.

Pretty much a matter of indifference to me (although I very much like
Smalltalk syntax). A little sugar in certain situations can be useful or
at least desired :) Remember this would be for the scripting
situations; one would use normal (or alternative) syntax for the usual
things.

> I believe we can achieve the same result by assigning [block] to
> variable and later [block] value:.

Well, since the syntax was pure sugar for these (and miniaml sugar, at
that) of course! ;)

> For serving pages, starting the image each time is really expensive.

Actually, this is what started the debate (if you track the thread
back). Lex claims (correclty I believe) that a script-focused image, with
scripts as image segmants, would load in time/space comparable to Python,
which, I would claim, is all we need. The OSProcesses goody gives us all
the nifty spawning stuff you could want. My small conceptual contirbution
was to permit text files to be compiled into image segmants on demand,
letting one cut out the Squeak IDE step.

Personally, that's not any kind of priority for me. But I could see where
some people would prefer it, and given that it's doesn't seem difficult or
even very tedious, I don't see any reason *not* to give it to 'em.

> Is
> there any thoughts on improving? I'm more from a C/C++/java/python
> background, but am willing to help.

See above and the rest of the thread. Lex has some timings somewhere.

> Thanks for taking the initiative,

Only for creating...the actual work of implementation I leave to the
masses :)

Cheers,
Bijan Parsia.