As a follow up to the previous conversation on KML QA; the patch we were concerned about has now now been applied to trunk:
- http://jira.codehaus.org/browse/GEOS-4597 (Thanks to Andrea and Michael Bedward for seeing this one through)
The work actually went pretty smooth, considering it was our third attempt, however I would like to ask anyone who uses KML to look for any regressions.
I was also thinking, even if the KML unit testing is light, perhaps we could use regression testing to check for any changes? What we actually want to to notice if the code changes KML output at all (so that we are sure that each change in on purpose). Regression tests could accomplish this just as well, or perhaps better than, regression tests.
The idea would be to generate out KML and perform a diff against known output; while it would not ensure that the generated KML is correct; it would detect any changes.