Fri, 08 Mar 2002 11:27:06 -0500
This is a multi-part message in MIME format.
Content-Type: text/plain; charset=us-ascii
Well ... so I am working hard to get unlimited resources.
For now, we more or less need a feature freeze, and a big Spring Cleaning
effort. We jumped some big hurdles based on coolness (so the Nightmare was
worth it, so to say - now wake up to reality).
BTW, I don't think the architecture of LON-CAPA is quite as bad as your email
makes it sound. lonc, for example, as messy as it was, has a very well defined
function and a very clean interface to the other components of the system.
There is nothing wrong with that. Ron could replace it by another lonc.
lond also has a very clean function to the outside, and should now be cleaned
up inside (for example, in the long command section, the "print $client"
should be replaced by "$output.=", and then a routine &sendreply($output)
called, which makes sure that there is one and only one "\n" in the
concatinated $output. Repeated code should go into subroutines. Stuff like
that - needs a lot of testing when done, since lonc and lond are so vital.
In summary, I am still confident that concerted effort of stabilization and
documentation can be highly successful, especially if a wider community of
programmers and developers starts to address this. We just have to resist the
temptation to implement the next cool feature.
> >>There is actually only one question: why did this ever work at all?
> Here's one possible answer: it was debugged to life. Since the system is
> so gargantuan I'm sure when it exhibited "correct" behavior it was assumed
> working; though never actually proven (mathematically or otherwise) to work
> correctly. When competing for financial resources it's no surprise to see
> something thrown together at first to prove how beneficial it could be so as
> to keep funding. If unlimited resources were available I'm sure parts of
> the project would be considered "throw-away" code or just a prototype of
> actual functionality and re-done later. However, this is obviously not
> possible at this stage of production. This is what one of my CSE profs
> would call a "Software Engineering Nightmare". I'm happy to be a part of it
> anyway though :)
Content-Type: text/x-vcard; charset=us-ascii;
Content-Description: Card for Gerd Kortemeyer
org:LITE Lab;DSME MSU
title:Instructional Technology Specialist
adr;quoted-printable:;;123 North Kedzie Labs=0D=0AMichigan State University;East Lansing;MI;48824;USA