130 likes | 224 Views
The ultimate solution for solving linear logical models. PRODUCT PRESENTATION. Content of presentation will not be published in advance!. T-76.115 Final Review. Rajoitteiset DE Iteration 6.4.2004. Working hours per person Quality metrics Quality assessment Software size Work practices
E N D
PRODUCT PRESENTATION Content of presentation will not be published in advance!
T-76.115 Final Review Rajoitteiset DE Iteration6.4.2004
Working hours per person Quality metrics Quality assessment Software size Work practices Tools used Agenda
Working hours by person Realised hours in this iteration Final summary
Quality metrics • Unit testing coverage improved, but still limited • 19 classes are tested automatically • System level testing framework available, number of test cases limited • 3 open bugs are actually missing new features, which have been marked as RESOLVED LATER into Bugzilla • CPU usage is not recorded • Epsilon is not fully working • The limits of the model are not binding the solver
Quality assessment Legend Coverage: 0 = nothing 1 = we looked at it 2 = we checked all functions 3 = it’s tested Quality: J = quality is good K = not sure L = quality is bad
Work practices • Documentation practice • proved to be a good way to ensure readability and consistency of produced material • Meeting practice • lack of regular ”mini-meetings” spoiled communication within project team • Design patterns • provided support to users familiar with the topic • Heuristic evaluation • was used to develop usability of GUI • peer testing was also partly executed as heuristic evaluation of GUI
Work practices (2) • Pair programming • practice was seen as pleasant way of working, but productivity is questionable • was used in most difficult modules, thus comparison between modules is biased • Unit testing • project did not contain maintaining the code and thus unit testing was not seen as useful • test cases were written too late • Automated system-level testing • building the testing framework took too much time • number of test cases remained too small
Tools used • Java, GNU Make, Javadoc, CVS and Bugzilla • ”old faithful” set of tools suitable for any project • Jlex and Cup • tools for handling the lm language • fulfilled their duties as expected • had limitations, but not a problem in this project • Trapoli • basically a moderate tool for time tracking, but had several shortcomings • with better guidance and manuals can be used also in future • maintenance of server needs to be improved
Tools used (2) • OpenOffice.org and Dia • good tools for documentation, especially on Linux • OpenOffice.org can be recommended • Dia had problems with incompatible versions • CCCC, Checkstyle, PMD, FindBugs • tools for evaluating the source code • were not used efficiently • had potential, in larger projects could be used more efficiently • Jetty • library for building Java Servlets • fulfilled our needs and seemed to be a good choice
Thank you for your attention! For any further questions regarding the project or commercial use of Lmodels, please contacthannu.kauppinen@iki.fi For information about research on solving linear-logical models or using Lmodels in your research, please contactjuhnu@cc.hut.fi Rajoitteiset DE Iteration6.4.2004