I’m working in a development environment of Mac OS X with MySQL, and a
production deployment environment of Redhat with Oracle XE. In the
course of setting up a new continuous build server that more closely
mimics our production environment, I got stuck on the following
strangeness on the build server (Linux, Oracle).
- rake fails
- rake test_units works
- rake test_functional works
We’ve reduced the problem down to a simple fault by using --trace … I
can see that “purge_test_database” does not happen between test_units
and test_functional when a complete rake is done, and the database has a
bunch of stuff in it from the unit tests, so it is not blank to start
for the functional tests.
So, I am confused. We have a complete match of Ruby and Gem versions
between development and build. Why is this behaving so differently? Is
this a platform difference (Mac vs Linux) or a database difference
(MySQL vs Oracle)?
I’m sure there is probably a simple work-around, but where should I