Chris, Well done. The automated testing should reduce the cost of providing quality significantly.
In answer to some of Tim's concerns re testing before committing: This could be partially addressed by offering automated tests for sandboxes too. If I have a complex fix, I could develop it in the sandbox, commit it, trigger automated testing, then merge back into the trunk. Christopher Schmidt wrote: > On Mon, Dec 17, 2007 at 05:26:06PM -0700, Tim Schaub wrote: > >> I like the idea (and if we were making a policy I would suggest) that a >> patch creator does three things: >> >> 1) make a ticket (more documentation than commit messages) >> > > I think that this is true, with the exception of things like: > > http://trac.openlayers.org/changeset/5439 > > Which I don't really think need a ticket. > > >> 2) run at least the tests you know are relevant in at least one browser >> 3) report what you did on the ticket >> >> Are you suggesting less? >> > > Nope, but I think that we're still liable to break trunk that way, and > I'd like to make it so that breaking trunk is found by a machine instead > of by a human. > > Specifically, I expect that as we support a wider range of browsers with > our tests (Opera is now as well supported as Firefox), we will find > things that break other browsers that we don't think to test... and it > lets us actually build a sane grid of what we support... and it lets us > ensure that we don't have to run around like chickens with our heads cut > off repairing Opera at release time. > > Right now, I think the requirement for 2) is ardurous because it's not > 'in at least one browser' -- it's "in at least FF + IE before it hits > trunk", and I think that's unrealistic long term as we add more > developers to the project. > > Regards, > _______________________________________________ Dev mailing list [email protected] http://openlayers.org/mailman/listinfo/dev
