Yes, this unittest has been painful to work with. Recently, I was trying to run the tests when I was offline and couldn't make this test pass. I spent some time trying to work around the problem but wasn't successful. I've disabled the test now: http://code.google.com/p/wave-protocol/source/detail?r=839d3a0de25d5f936309823711a5759d8847a586 It's possible that we should delete the console client and tests altogether, because they're hard to keep working as we improve the client-server protocol, e.g., I think they're the only reason we haven't deleted the hard-to-maintain index wave yet.
On Sun, Jul 17, 2011 at 4:44 PM, Andrew Fleenor <[email protected]> wrote: > I recently cloned the Hg repository to try wave out. When I compiled, most > of the ConsoleClientTest unit tests errored out with NullPointerException's, > specifically, all those that called ConsoleClientTest.connect(). When I > eventually tracked the problem down, it was because the ClientBackend was > actually trying to resolve the bogus test server "host.com". > this.client.backend silently failed to be set (since the IOException was > caught, with the only handling being a debug print), causing the > NullPointerException later on in ConsoleClientTest.connect. When I added > host.com to /etc/hosts, it all went smoothly. > > For that matter, it ran smoothly when I just disabled the tests, which > is how I got my first look at the new UI. > > Obviously, it would have been nice to know about this before I started. > Updating the documentation would be good. Fixing the error handling would > be better, but I would understand if that's not a big priority, given the > nature of the problem. In any case, thanks for keeping this project alive. >
