   
Summary
|
Use Frames
Support.com
|
zipRealty.com
|
Pharmatrak
|
Thomson & Thomson
|
Wellington
Thomson & Thomson
|
BBN
|
Fidelity
|
Teradyne
|
Sapient
|
GCC
|
Digital
|
MIT AI Lab
|
Temple
Thomson & Thomson
Internet, Copyright & Trademark
Quincy, MA
May 1997--August 1997
Perl, Netscape Publishing System, Oracle, Java, Unix

Thomson & Thomson is a world leader in trademark and copyright
research. With sister companies in Canada, Belgium, and Japan,
Thomson & Thomson provides one of the most current, accurate,
and comprehensive collections of trademark information available.
Services include traditional domestic and international copyright
and trademark filing, searching and reporting, on-line World Wide
Web products for self-conducted screening, and watchdog monitoring
of newly published marks to warn of potential infringement.

The SAEGIS
product is Thomson & Thomson's on-line trademark and copyright
search engine, allowing customers to perform their own customized
queries and reports. Based on the Netscape Publishing System (now
known as the PublishingXpert),
the system allows corporate users to login, accrue individual
charges based on a variety of SAEGIS products, and bill to the
client's parent company.
Project responsibilities revolved around the billing aspects of the
system. While the Netscape software provided a solid framework
from which to build a billing system, much custom work was
necessary to support the flexible pricing structure needed.
Furthermore, SAEGIS required special processing such as corporate
discounts, taxes, and importing to MACSS, Thomson & Thomson's
legacy system, which were not supported by the Netscape system.
The original procedure for closing and exporting corporate billing
statements had several problems. First, the overall process of
collecting due statements, preparing them for closing, and then
actually processing them was very manual, requiring close scrutiny
whenever run. Also, its work around for a serious Netscape bug in
closing statements was tricky and error prone.
The other major problem was the actual program for processing
statements. It required shutting down the entire Web server in
order to run properly; was cumbersome and very hard to modify for
new features; and had very complex and risky rollback handling in
case of errors.
The first task was to completely redesign and rewrite the statement
processing program to overcome these difficulties. The new program
used a very different approach to closing statements and
interacting with the necessary Netscape software, neatly
circumventing the need to shut down the Web server and narrowing
non-committal of database changes in case of an error to a single
database rollback statement. Given the benefit of a redesign, the
new program was also overall much more modular, and thus easier to
maintain, debug, and upgrade.
The next task was to examine the manual approach of collecting due
statements and setting them up for processing. The modified
procedure was completely automated, and much more cleanly and
safely avoided the Netscape bug. It was also more modular and
flexible, allowing individual components to be easily tested,
modified, and rearranged.
Finally, rewriting the program in Java was evaluated since much of
the SAEGIS system was starting to be rewritten in Java. While
initially implemented directly on top of the Netscape schema,
future plans for SAEGIS involved an object-oriented, CORBA-based
layer through which all components---rewritten in Java---would
communicate.
|