This is partly personal style, but there is a good reason behind it too. For one thing, if you have one big method, and the belongs_to test breaks, it precludes the later tests from being run. So one failure can mask another. For another, using smaller methods lets you name them more descriptively.

Great post though - thanks!

class WineTest < ActiveSupport::TestCase
# omitted unneeded :varietals_wines fixture
fixtures :wines, :wineries, :varietals

def setup
…
@bordeaux = wines(:bordeaux)
end

def test_relationships
…
# has_and_belongs_to_many :varietals
assert_equal(3, @bordeaux.varietals.count)
@bordeaux.varietals.each do |var|
assert(var.wines.include?(@bordeaux))
end
(Varietal.find(:all) - @bordeaux.varietals).each do |var|
assert(!var.wines.include?(@bordeaux))
end
assert_equal(”Cabernet Franc”, @bordeaux.varietals.sort_by {|v| v.name}.first.name)
end
end

In my example, the idea is that I’m testing two things to make sure they are the EXACT value I expect. In your example, the idea is that you don’t care what value they have, just that they are equal. If @merlot.winery_id and @merlot.winery.id both have a value of 4 then my tests will fail and yours will pass.

The way you decide to write your tests will determine if that distinction makes a difference or not.

Could we just write this?
def test_relationships
# belongs_to :winery
assert_equal(@merlot.winery.id, @merlot.winery_id)
[...]

Thanks
Denis

Ler este artigo em português
I’ve been looking for an elegant way to test relationships using Rais, and think it’s interesting how so few people really care about it. Most simply say that it’s up to the framework to test it’s o…