On Wed, Dec 17, 2008 at 6:38 AM, Vincent Siveton <[email protected]>wrote:
> Hi Kevin, > > 2008/12/16 Kevin Brown <[email protected]>: > > [SNIP] > > > Out of the box, some are going to fail because they depend on the storage > > mechanism (activities / app data) being writeable. If you want those to > pass > > you have to wire up persistence to a writeable store. The default stores > are > > read only. > > Thanks for the clarification. > > > Most of the warnings are bogus IMO -- the compliance tests attempt to > define > > error conditions that aren't in the spec. I've brought this up with the > > But tests (ie PPLX00.7) speaks about error code: > FAILED: (got 'badRequest'), expected 'unauthorized' This one may be because of persistence wiring. When I ran the test on Orkut this passed without issue, and Orkut is running something pretty close to what's on the release branch. > > > compliance test maintainers on several occasions but they disagree on > > interpretation of "not specified". If somebody wants to write patches to > > make shindig "compliance gadget compliant" instead of "spec compliant", > > though, I guess that's OK. > > So, which tests are purely gadget or spec compliant? So far as I've seen, it's just some of the warnings that have questionable interpretations of what the spec states. The error conditions all seem to be accurate. If you run the compliance gadget and only concern yourself with errors, you'll probably be OK. The warnings require close inspection. The most common example that you'll find are that some methods throw exceptions on errors, and the compliance gadget doesn't expect an exception to be thrown. The spec doesn't state what the error condition is, so neither is necessarily right or wrong. The spec really should be clear on these issues, but for some reason they keep getting overlooked. > > > Cheers, > > Vincent >
