[Prev][Next][Index][Thread]
Re: Believing in what you sell...
At 9:31 AM -0800 1997-02-15, Kelly Murray wrote:
>This Franz engineer knows the solution to the mystery ;)
>
>I have studied CL-HTTP quite a bit and understand it enough to have
>incorporated our AllegroStore CLOS object database system into it, and
>also developed a timecard application which ran internally within Franz.
>This CL-HTTP timecard app was used daily by Franz employees
>until recently.
This is the first that I heard of these. Not even a bug report despite
the fact that you and Jim Veitch were on the bug-acl-cl-http for a long
time. Too predictable. But, perhaps you consulted directly with Olivier
Clarisse, who put so much effort into the ACL CL-HTTP port and
made it very popular.
>
>The timecard app is now running under a new, entirely different Web
>server which I developed in Lisp for Franz.
Did you learn anything from CL-HTTP?
Can you explain any concrete deficiencies that you feel should be
corrected?
Do you feel you need put anything back into the community?
What will you do with the old cl-http code you won't be needing any more?
>This server, which we are calling Charlotte, focuses
>on simplifying the task of creating and managing a website. As such, it
>has a different purpose and a different target user base than CL-HTTP.
What is different about the purpose?
Does it provide http service?
Can you enlighten us on some of your design issues?
Do you implement HTTP 1.1?
Are you prepared track changes in HTML, HTTP and related standards?
Will you distribute sources?
We have a large publications and collaboration system that uses
CL-HTTP for its web interface. It has a large amount of database code.
We clearly differentiated Web code from database code for modularity reasons,
but we specialize class into database entity/objects.
We were careful not to wedge the server into particular applications
or databases in order to maintain generality and flexibility.
We find that putting slow databases, and for that matter disk drives, into
http transactions leads to poor performance.
How do you address this issue?
Finaly, did you or Franz consider how you could have expended your
resources more productively by doing some innovation, e.g. multiprocessor
support?
References: