Experiences with Publishing on the WWW
Evi Nemeth
Publishing on the web is so easy that maintainability,
network impact, and information content are often overlooked
in favor of whizzy buttons, fancy graphics, and colorful
backgrounds.
This paper includes experiences and lessons learned in several
publishing ventures at the University of Colorado:
- All Computer Science Dept information for prospective
students, current students, and the public;
- The semi-automatic conversion of the University Catalog
from the quark express SGML language to HTML (with
lots of help from Perl);
- A jobs database for employers in the area to register
their computer science related jobs;
- A source code tracking system for software installed
from the net or written locally;
- A login/password/card-access system for new students to
get accounts and access to labs;
- A graduation audit system where students can query
departmental records and requirements to plan their
courses.
Where existing materials were converted to HTML we
concentrated on organizing them in such a way that
non-technical secretarial staff can maintain them. The
catalog conversion project required a series of Perl
scripts to correct both the formatting codes and the
differences in host architectures. Many hidden errors
in the original catalog were exposed by the translation.
The final result was about 95% complete.
In both of these projects, reasonable use of network bandwidth
was a key factor in both organizing the sizes of chunks of
information and in the use of pictures and images.
Security of cgi-bin scripts was an important issue in the
last two projects. Their use is too new to know if we were
successful in our approach.