On 11/09/2007, Rafael Schloming <[EMAIL PROTECTED]> wrote: > Robert Greig wrote: > > On 11/09/2007, Carl Trieloff <[EMAIL PROTECTED]> wrote: > >>> We now have an automated build running on a reasonably powerful box so > >>> we will be hammering *all* the tests going forward. > >>> > >>> > >> We also build all on every checkin of M2 + trunk (across the > >> architectures). I hope to > >> send the failure to the commits list. > > > > In that case I don't understand why at least one of Rafi's bugs was > > not detected sooner? > > > > Were they not failing on the machine that build was running on? > > All of my bugs were intermittent failures due to race conditions that > for some reason my dev box is more prone to than others. My current > theory is that I have the right number of processors (2) plus most of my > memory eaten up by firefox when I run the tests. ;) > > --Rafael
I haven't seen the TransactedTest ever fail and I use a Core2Duo 2Ghz machine with 2G of RAM. One of my colleagues with a Dual 3.2Ghz machine had real difficulty reproducing the bug even with the delay. We do both have windows on the machine so perhaps that affects things. I have raised a JIRA for the CombinedTest that of late has been less intermittent that it had in the past. I do think some of our tests are overly complicated and poorly documented as a result it is difficult to know what they are testing. -- Martin Ritchie
