...
Engineering the release process
- Converting manual UATs to automated functional tests
- Running these automated functional tests nightly
- Against a CAS server build deployed into an SSL'ed Tomcat nightly
- With an SSL'ed helper servlet
Where to do this
Andrew suggestion that we start but just wandering off into a dark corner, close the door, and just "make it happen" by an unholy conglomeration of cron jobed shell scripts running Maven on portaldev.its.yale.edu, because we have root and can probably make it work. Then, once we have the immediate value of nightly runs of the tests, revisit how to move this onto Clearinghouse.
Scott's, others' opportunity to talk Andrew out of it by suggesting that doing this on Clearinghouse first pass doesn't add that much work.
Resources
- Yale: Drew and Andrew: finish producing automated functional tests for all UATs persuant to less heroics for release. Timeframe: After immediate Sakai heroics but before 3.0.2. Expected delivery in 3-5 weeksearly October.
- Yale: Drew and Andrew: automate nightly build process to actually run all these functional tests against a CAS instance deployed into an SSL'ed Tomcat. Timeframe: After immediate Sakai heroics but before 3.0.2. Expected delivery in 3-5 weeksearly October.
- Yale: Howard: Yale Shib pilot project, fronted and consumed by CAS. Immediate term (over the next month). Will have deliverables suitable for inclusion in CAS 3.0.x (fully backwards compatible) and 3.1.x (incremental evolution).
...
- Reaffirmation of policy on urgent issues: critical bugs, security issues for which a known fix: get them fixed and out in an ASAP release.
- Barring such a catastrophe, reasonable estimation of 3.0.2 timeline and content
- Realities of academic term starting: realistically, no release sooner than Sept 15th or so, right?
- Goals for 3.0.2?
- Functional testing
3.1.0
- Evolution that cannot be accomodated in 3.0.x
- Howard et al. Shibboleth integration?