[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

scheme/dylan (was Compilation on SPARC)



> From: Scott Schwartz <schwartz@roke.cs.psu.edu>
> Subject: Compilation on SPARC
>
> rpg@cs.tulane.edu (Robert Goldman) writes:
>
>   why hasn't anyone else?
>
> My (quite possibly incorrect) observation is that, for most people's
> purposes, an interpreter adequate.  Most people who use scheme seem to
> be either
> 
>  1) doing research on programming languages.  unless they are researching
>     optimization techniques there is no point in writing a compiler at all,
>     and if they are, only one platform is necessary.
> 
>  2) teaching intro cs classes
> 
>  3) using scheme as an embedded command language.
> 
> Only a very small number of scheme developers seem interested in using
> scheme do to actual systems and application programming.  (If I'm
> wrong about that, post lots of messages here!)  I wonder if an
> advantage of Dylan will be that its user community will have different
> goals than the scheme community.

I agree with this.  There should be a language suitable
for programming language research and teaching the issues of
programming language design.  Such a language needs to be free to
change, and to adopt constructs which may not be efficient at
the present time, in the view of some (this is how I view scheme's
numerical system!).  These characteristics are the opposite of those
desired of a 'real' or 'utility' language, i.e., stability and efficiency.

Dylan is aimed at developing 'real' applications, so the burden on 
scheme is now reduced.  I hope Dylan catches on.  Mail to info-dylan
from Rob MacLachlan indicates that a high performance Dylan implementation
is being developed at CMU.  Given the CMU role/experience with Common Lisp
and the possible efficiency of Dylan,  I am greatly looking forward to this.

Although CL, scheme, and dylan should probably be considered as 
having more differences than e.g. C and Pascal, thanks to their macro
systems it should not be too hard to port moderate amounts of
code to Dylan when it becomes available.

j.p.lewis/nec c&c research