[plt-scheme] ipc among mzscheme processes

Raymond Racine <rracine at adelphia.net> writes at 07:33 11-May-2003 -0400:
> I ran across the DTYPE protocol awhile back.  But this wheel has been
> chipped from stone many times in ancient BX times (Before XML)

Ah, thanks for the reminder of FramerD.  I'll have to take a close look
at what he's doing wrt IPC.

> - MIT Media Labs developed a great deal of software in Lisp.
> - There was shock and dismay in that no one seemed to use it.
> - Reason: they were Lisp, the rest of the world was C.
> - DTYPE protocol was to expose Lisp based software to C.

For FramerD, the fact that it used a Lisp language very likely was the
main barrier to external/industry use.

However, as a general comment on Lisp and technology transfer, I don't
think that language is generally the main barrier to people adopting
software developed in a research environment.

Even during dotcom investment mania in the mid-to-late '90s, when the
Media Lab used Java for the vast majority of projects, and there were
countless industrial sponsor relationships, I don't recall any research
code actually being used outside the ML.  Even the startups I know of in
which a business plan was developed in parallel with the MS thesis, I
don't think the research code was used.  This wasn't just a question of
prototype vs. production code; the student startups, while ostensibly
addressing the same problem areas as the research, often didn't actually
use the student's own research approaches.  An oft-repeated quip by one
of the more successful student entrepreneurs expressed the reality best:
"first thing you do, throw out all the AI."

The main problem I see with doing research prototyping in a Lisp is that
in a post-Java world it's hard to find slaves -- er, research assistants
-- who can quickly become competent Lisp hackers.  Hopefully HS programs
like TeachScheme and college intro CS courses using DrScheme will help.

-- 
                                             http://www.neilvandyke.org/