[plt-scheme] llvm and all that

On Aug 10, 2006, at 9:16 PM, Ray wrote:

> I'm sure some of the PLT core have must have considered the  
> possibility
> of leveraging Larceny in some fashion.  What are the major show
> stoppers?

After four years of raising and spending some reasonably large amount  
of money (you could buy a nice house in Boston or Sillicon Valley for  
that) on bridging the gap between PLT Scheme and Larceny, I have no  
hope that I will see it in my life time. Perhaps you're younger and  
may benefit from the fruits of a 3-decade undertaking. The effort  
involves moving the two to a common neutral platform and lifting  
Larceny's functionality from basic to, well, more than basic. The  
common platform is .Net and there still isn't enough scriptability of  
the library to even think of running MrEd-style programs.

In a nutshell:

PLT Scheme supports a large array of functionality (in the core and  
in the libraries) that Larceny just doesn't get remotely close to and  
I don't see how they can. Example: When we write joint NSF grants I  
can use a 10-line PLT Scheme script to extract and create a  
bibliography. In Larceny, it's impossible as is and to get close to  
the same functionality with major shortcomings takes a couple of pages.

At the same time, the Larecenists will tell you that PLT Scheme is so  
badly designed internally that it's causing physical pain to look at.

So that's where things are standing now.

-- Matthias