Transcript Final Project Presentation [61 KiB]

Distributed Software Development
Web Project
Final Presentation
2016-05-24
1
Project summary
• Things not done:
–
–
–
–
Multilingual support
PERT diagram generation
Gantt diagram generation
Project baselining & overview
• Things done differently than planned:
– Enterprise Java Beans middle tier overhead
framework code cost us too much time
– JDBC connectivity was the solution (a late solution,
unfortunately)
2016-05-24
2
Development obstructions
•
Enterprise Java Beans:
– A lot of overhead framework code turned a single SQL query into 1kbyte
of code
• DataObjects classes
• Framework made of entity beans, session beans and their interfaces
(automatically generated by development software)
• Class instantiations in JSP
• Difficult implementation and debugging
•
– Huge system requirements (512 Mb RAM and at least a Celeron
processor at 1GHz for comfortable operation) - too much for Orson,
obviously :)
– Solution: JDBC connectivity
PostgreSQL:
– INSERT statement on PostgreSQL 7.3 (used locally) had a different
format than 7.2.1 on Orson (always pay attention to backward
compatibility :))
– GRANTING access rights to database users might not be as simple as
you think it may be (at least to spoiled Informix brats like us :) )
• Each and every table, sequence generator along with other database
elements used need to be declared separately
– Creation of sequence generators made for serial datatypes also differs
on these two releases
2016-05-24
3
Development obstructions II - The saga continues :)
• Application schema:
– Overambitious for the time frame of the past semester
– Huge dataflow (32 multiply interconnected relations) required a more
sophisticated and gradual approach to the user interface problem
– User interface schema thought completed and needed to be changed
once the database tier became operational and once the application
started crunching data
– Changes made by developers were interdependent and introduced
frequent changes in everybody's work
– Frequent additions and changes to the database schema
• Planning:
–
–
–
–
Could have been done better
Too much time spent on EJB, sometimes it's good to give up early
Two weeks of team member's idle time
Too much time spent on efforts that turned out to be futile
2016-05-24
4
Development obstructions III - Above and beyond :)
•
•
The final outcome
– 12 hours of daily work (on average) during the past two weeks most of
which took place in my apartment
– Caffeine poisoning
– Sleep deprivation
– Monitor radiation overdose
– Some parts of the application are left unfinished and not tested properly
The moral of this story
– Negative experiences
• Analytical approach to project management is far better than intuitive one
• Careful workload evaluation is essential
– Positive experiences
• A somewhat extensive knowledge of:
–
–
–
–
–
2016-05-24
JSP
PostgreSQL database administration
User interface design
Including the unfortunate Enterprise Java Beans "field trip"
teamwork
5