[info-mcl] Pathname and Unicode in MCL 5.1

[Ron Garret]

On Oct 2, 2007, at 1:58 PM, Toomas Altosaar wrote:

>> As far as reuiniting OpenMCL and MCL, I know that is not a
>> possibility. I would really love to have Fred on OpenMCL and some
>> reasonable path for porting gui code from MCL. If we all chip in
>> $1,000 maybe we could hire someone to do this (hahaha...).
>
> I think Sean has hit upon an important issue here: whether a working
> version of MCL costs 750 USD or 1750 USD (or even a bit more) makes
> little or no difference compared to the costs of spending
> months/years porting code to another Lisp, another OS, or even
> abandoning Lisp altogether and moving on to another language.
>
> There is obviously an acute demand for MCL to continue, so why are
> there no offerings for the development and sale of an IDE from
> Digitool or Clozure or someone else?
>
> Or is it that what Digitool developed, i.e. Fred, so unique and
> powerful, that no one else can replicate it? So far it seems so.

Just because no one has replicated it doesn't mean it's not possible,  
just that no one capable of replicating it has deemed it worth their  
while to do so (yet).

For the record I retain a keen interest in porting and/or replicating  
MCL to/on OS X.  I don't have enough technical expertise to  
contribute directly to this, but I do have financial resources at my  
disposal.  The limiting factor in getting this done is not capital,  
it's getting someone to come up with a coherent plan on how to make  
it happen that makes business sense.

However, my current understanding of the situation is:

1.  The 32-bit Intel architecture is too register-poor to admit a  
reasonably efficient port of MCL without an extraordinary effort.   
Since 32-bit is probably legacy technology anyway, it's probably not  
worth putting forth this effort.

2.  A 64-bit compiler is in hand, but there are no 64-bit interfaces  
to the Cocoa library and won't be until Leopard is released.

Therefore

3.  We are all holding or collective breath for Leopard to be  
released, whereupon the logjam should break and a new version of MCL  
should appear.

The best strategy therefore seems to be simply to wait.

If I've gotten any of this wrong please set me straight.

rg

P.S.  One technical strategy which occurred to me as a way to break  
the logjam sooner is to write a thin 32-bit client in C that provide  
a minimal user interface (i.e. nothing more than "move the cursor  
here, insert a character there, report a mouse click", like a little  
X server but specifically tailored to serve Fred) and communicates  
with 64-bit MCL over a socket.  I throw this out just in case it  
makes someone with more hacker mojo than me to say, "Aha!" and build  
this in the next 48 hours.