[Larceny-users] do you want YouWantItWhen ?

[Ray Racine]

For me, basic module/library support offering simple import/export
capabilities (i.e. straighforward namespace scoping and encapsulation)
which does not adversely impact the efficiency of compilation is pretty
important. 

I would go so far as to say it is the issue gating the timing of when I
would switch to Larceny as my primary Scheme.  I've been patiently
watching progress and am not asking the implementers to do anything rash
here.  Anecdotal observation from afar, on occasion, does indicate the
implementers do have a plan and are progressing.

If the goal is here is to determine the desire of the user community for
module/library support with some semblance of a sane deterministic
expansion at compile time ( I would not go as far as to say intuitive.),
then yes, its pretty key for me.

"You want it when", or equivalent works for me.

Ray

PS. I have seen the recent R6RS creation lib dir in SVN based on Van
Tonders renaming approach. My fingers are crossed.

PPS.  I did do a hacked up attempt at this myself.  I made no effort for
let-syntax support, did only naive chain following etc... used Van
Tonder's Dec/Jan release of his system.  Separate compilation was to a
xxx.lep file, for [l]exically [e]xpanded [p]rogram.
 http://en.wikipedia.org/wiki/Leporidae

This was before Andre did the more recent overhaul of his system which I
have yet to look at in any detail.

Given a specific library/module, I did naive you-want-it-when binding
all module identifiers, and imported identifiers into a Larceny 1st
class environment and toggled the REPL into this environment.  

loop: 
 play with library under development, 
 make changes in emacs, 
 toggle out to Larceny repl, 
 repeat compilation of library to a new environment, 
 throw old library environment away, 
 loop.  

The goal was to emulate PLT Module Mode.  It was too much for my pea
brain and remaining life span to finish.  And I needed to be working on
other things, so I remain in PLT, which isn't so bad at all modulo
performance.

To really make the whole thing work one would need to have a
designed-for-Larceny implementation of Slime.

Now that would be nice.  Maybe some of us user-community leaches could
band together and make a run at this ...



On Tue, 2007-09-04 at 00:01 -0400, Felix wrote:
> Will-
> 
> On Sep 3, 2007, at 8:00 PM, William D Clinger wrote:
> 
> > Felix wrote:
> >> The recent discussions with Dan Muresan led me to think about
> >> implementing SRFI-55 in a manner that imports syntax at compile-time.
> >
> > I intend to implement a prototype of ERR5RS for the next
> > public release of Larceny.  Since this might solve the
> > problem and is likely to influence our solution in any
> > case, I'd suggest you hold off for now.