On 03/05/2008, at 3:20 AM, Jeremy Stell-Smith wrote:
and don't forget, developers make mistakes...all the time. if they
didn't there would be little point to a continuous integration
system in the first place. what if I change a migration, then
"forget to run it" before checking in. this is exactly what ci
should protect you from, however, in your case instead of just
failing, ci will go into an error state that the only way to fix is
to log onto the box.
no good.
though it'd be nice for this *not* to happen. Surely there's a way
around it?
I haven't used CC.rb in some time, but the way I used to have it set
up is that the production DB would be copied for CC.rb to test
against, then migrations were run, then the tests. If the DB was
borked after one check-in the next had a chance of fixing the
migration as it'd rebuild the DB from production and rerun the
migrations.
I don't like the idea of assuming your CC.rb schema is the same as
production simply because you're rebuilding from migrations each time.
-- tim
_______________________________________________
Cruisecontrolrb-users mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/cruisecontrolrb-users
