[Prev][Next][Index][Thread]
Re: Believing in what you sell...
At 10:30 PM +0000 1997-02-13, Simon Brooke wrote:
*If* I manage it, I'm going to
>write some 'for dummies' documentation, because I find the existing
>documentation has too high an assumed knowledge threshold for me to even
>begin to understand it (hell, I've only got twelve years of professional
>LISP and three years of professional Web service).
Don't worry in another decade it will all be transparent! :_)
Actually, it would be great if you noted all stumbling blocks that
you encountered so that the docs can be improved. Practically,
all the existing docs have been written by the developer (that's me)
as I go in spare cycles (which are few).
Apart from how to get started, we probably need a document that transfers
the conceptual model concisely (the is the www94 paper). But, there
is probably a gap between the papers and the code level documentation.
Some knowledge of how HTTP works is usually helpful. Most of our interesting'
applications are in systems that we do not hand out, and so the example
code is limited. More examples would be good. Forget not that there
several tutoring systems written atop cl-http, and these coud be
used to provide really nice documentation for new users.
Franz doesn't run the server because their engineers claim they can't understand it.
The trot around the world explaining this themselves (!). Perhaps their development tools
are not so good or their engineers not so clever. Who knows? One of the modern mysteries.
At any rate, some people can't stand free, high quality, open code developed by users
that promotes their core business in a critcial way. Go figure.
Digtool Applied Technologies has a product built on cl-http.
Digitool, Harlequin, and Symbolics wizards think it is easy. Go figure.
The vendors might find the mail hyperarchive a useful extensible tool
that they could run on their sites.
As far as I can tell, code doesn't come much easier than this, but it does
presuppose full mastery of common lisp.
The major usage of CL-HTTP is on intranets where more sophisticated technology
is deployed. NEXOR indexes servers that you tell it about. (it may also be buggy)
People with public CL-HTTP servers should type them in. We run a number of these at MIT.
The media lab does as well. ISI and CMU use it as do numerous other universities in
research contexts. There are a number of government users, one of which is
one of the highest profile sites.
In any event, you make some good points. If the lisp vendors were wider
awake, we would be seeing more lisp activity in general. I think the demise of
Symbolics slowed progress down because there was nobody to copy anymore. However,
I think we are now at groundhog day in the AI winter, and you will see
a major revitalization of lisp and dynamic languages over the next five
years, driven largely by the exigencies of electronic commerce and
fast, cheap, memory-laden commodity processors.
The wonderful thing about lisp and CL-HTTP at the moment is that we have
the world's most talented users. I'd trade that any day for the teaming hoards
using Apache. What we need are more intelligent contributions from the user
community to further enhance the environment. If we want teaming haords, we'll need to
put a trivial UI on the server and distribute binaries for all platforms plus add some
unique yet catchy functionality. Of course, this is not something we at MIT
are prepared to do in a research context, but we hope users or commercial
versions will do.
Follow-Ups:
References: