[Gardeners] In a Box options

[Peter Seibel]

On Dec 14, 2005, at 9:56 PM, josh giesbrecht wrote:

> David Aue wrote:
> In hindsight I would like to make the
>> following suggestions for a LISP in a Box 2.0.
>>
>> 1) Full power and free tools
>>
>> 2) Most people run Microsoft Windows
>>
>> 3) Packages make programming fun
>>
>> 4) Learning a new and totally unfamiliar editor sucks
>
>
> Let me just say: yes, yes, yes, and again yes.
>
> I had a LOT of my Lisp enthusiasm sucked out of me simply because  
> of all
> the hassles and limitations (both real and perceived) of wanting to  
> set
> up a free Lisp environment to use in Windows.

So as CL Gardeners dedicated to doing work that improves, however  
slightly the current situation, we have two choices here: we can  
attack the real or the perceived difficulties of getting a free Lisp  
environment that works on Windows. At the moment the situation on  
Windows for free (as in beer and/or speech) Lisps is this:

   1) You can use CLISP as long as you don't need multithreading and  
can live within the constraints of its GPL license. (Which doesn't  
quite require you to GPL all your code but does in some  
circumstances.) This is the best solution if you want a free as in  
speech Lisp.

   2) You can use Franz's trial version of Allegro as long as you  
don't mind renewing your license every 60 days and don't plan to do  
commercial work with it. Only free as in beer.

   3) You can use Corman which is not quite free (as in beer) but is  
quite inexpensive. Not free as in speech. I think it also has some  
weird conformance idiosyncrasies but I've never really used it so I'm  
not in a good position to say what they are.

   4) SBCL is getting quite close to having a Windows port. When it's  
ready we'll have a free as in beer Lisp under a BSDish license which  
will be better for folks who don't want to wrestle with the GPL  
implications of CLISP but who don't want to deal with negotiating a  
license with Franz for whatever thing they want to do beyond trial  
usage. It won't, as I understand it, provide a free (as in beer) Lisp  
with multithreading without another fair bit of work.

So folks who are inclined to tackle the real technical problems  
standing between Windows users and Lisp adoption, the best bet would  
either be to help out the SBCL effort or (if you think threads are  
important) to try and figure out how to help the CLISP guys add  
multithreading to their implementation. The latter seems like a  
pretty big challenge but if there are any hot-shot C programmers out  
there with experience working on VMs or low-level OS scheduler type  
code, maybe you want to take a crack at it. As for the SBCL port, I  
believe the author of that port is on this list so maybe he can say  
more about it but I'm pretty sure he's looking for volunteers to help  
out.

I'm not sure what exactly can be done about the perception problems  
except possibly to point out that with CLISP you get a cross-platform  
solution similar to what you get with Python or Perl. Or at least  
those languages as of a few years ago before they had threads. (Do  
folks really use threads in Perl and Python these days? I left Perl  
hacking before threads were really soup.) Are there other perceived  
difficulties that we could do something to mitigate?

-Peter

-- 
Peter Seibel           * peter at gigamonkeys.com
Gigamonkeys Consulting * http://www.gigamonkeys.com/
Practical Common Lisp  * http://www.gigamonkeys.com/book/