Attending: Jeffrey Barnett, Franck Borel, Demian Katz, Till Kinstler, Daniel Lovins, Preetha Rao
Due to problems with its PHP plug-in, our Sonar server is not very useful yet, though its code complexity and copy-and-paste detection functionality should prove useful. Thanks to Phing, we are now using Hudson to track code standard compliance and unit test coverage.
Right now, the value of our unit tests is minimal because we do not have good tests for code which relies on external services (i.e. Solr, MySQL, LDAP, etc.). Since this forms the bulk of VuFind's code, our coverage is low.
We discussed possible solutions to this problem:
Further feedback on testing issues is welcome – please post to the vufind-tech list if you have thoughts on the issue.
Demian has begun cleaning up code to better comply with the PEAR Coding Standards. Our compliance is visible in Hudson's checkstyle area. If you add new or maintain existing code, please check its compliance in Hudson to ensure that you are within the standards. Feel free to contact Demian (demian DOT katz AT villanova DOT edu) if you need him to trigger a Hudson build so you can see the latest results – builds are not yet happening automatically.
We briefly discussed Demian's problem with figuring out how to reconcile long PHPDoc parameter descriptions with the PEAR standard's line length limit (i.e. is there a way to do multi-line PHPDoc token values?). Jeffrey will investigate.
We discussed the VUFIND-304 JIRA ticket. See comments on that ticket for background information. Demian will attempt to put together a patch for discussion at the next call (or the call after that, if he is unable to finish in time).
Daniel asked if scope notes can be added to VuFind's Terminologies recommendations module. This should be possible by extracting appropriate data in the getRelatedTerms() method of web/sys/WorldCatUtils.php and displaying it in the templates found in web/interface/themes/[your_theme]/Search/Recommend/WorldCatTerms.tpl.
Till asked if it would make sense to attempt to parse MARC records as XML if treating them as binary fails when initializing the MARC record driver. There was no objection to adding this flexibility to the code.
The next call will be Tuesday, November 30 at 10am Eastern Standard Time (GMT - 5:00).