You can avoid it by running the tests manually
ruby -Itest test/unit/some_test.rb
It is the rake task which does the test db recreation (you can run it manually like so)
rake db:test:prepare
But my suggestion is that you're doing it wrong.
The general idea in testing is that you know the state of the database, and therefore know what to expect from a function.
eg.
test "search_by_name" do
expected = User.all.select{|u| u.name =~ /arthur/i}
assert_equal expected, User.search_by_name("Arthur")
end
is a fine test
however, if you don't know the state of the db, how do you know there is an arthur?
The test above would pass in three bad cases;
- there are no User records
- all Users are called "Arthur"
- there are no users called arther.
So its better to create a false reality,
where we know the state of the database.
We want;
- at least one user with the name "Arthur"
- at least one user with the name "Arthur" as part of another word
- at least one non-Arthur user.
a better test, assuming the db is empty, and using factory girl, may be.
test "search_by_name" do
expected = [
Factory.create(:user, :name => "Arthur"),
Factory.create(:user, :name => "MacArthur")
]
not_expected = [Factory.create(:user, :name => "Archer")]
assert_equal expected, User.search_by_name("Arthur")
end