On Mon, 24 Aug 2015 12:01:53 PM Andrea Aime wrote: > On Mon, Aug 24, 2015 at 11:28 AM, Brad Hards <[email protected]> wrote: > > After some work on CITE, I'm unhappy with the effort required to do a > > local > > test (since you have to set up the test data). So I was going to suggest > > that > > we modify the "normal" release build to include a working test environment > > such that tests will pass against a routine build without needing to > > modify > > the data directory. > > Interesting idea... with some gotchas. > > The first one is that the CITE tests cannot be run against shapefiles due > to their attribute naming and data types > limitations... thus, we would have to ship H2 data stores with the right > data for the WFS compliance tests, instead, > and as a result, bundle the H2 store as a core one. > I'm not against, but would like to hear other devs opinions. I was thinking perhaps property files might be an option. > Another one is that the CITE tests expect their given data set, and I'm not > sure what happens > if other layers are made available as well. I think that would be a bug :-)
> We could try to use separate workspaces and use workspace specific > services, but WFS 1.0 > needs two separate workspaces, so at least that one would have to be run > against global services. OK. Will investigate. > Then there are the WCS tests, which fail if there is more than one version > of WCS deployed, > one has to selectively pull jars out of WEB-INF... again, we could add a > new feature to selectively > enable/disable WCS versions, would be nice, but to run the different tests, > one would still have > to unckeck the other versions... I guess I'd rather fix the CITE tests > instead. OK. > For CSW tests, there is a gotcha that we would need to run against the > "basic store" instead > of the internal catalog one, so again a classpath manipulation. We could > make this configurable > again, or try to work on a composite store that allows more than one > backend store to be > used (would be useful too). OK. > And then there are the other new tests that we are hopefully going to add, > not sure what > their requisites are, believe for example that WCS 2.0 has its own set of > expected data. Maybe a separate workspace could be an option here. > Hum... cannot think of other issues off the top of my head, but the ones > above are more than > enough imho :-) Maybe we could have a progressive delivery on this. Say start with WMS 1.1.1 and 1.3, and work up? Brad ------------------------------------------------------------------------------ _______________________________________________ Geoserver-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-devel
