Test suite assertion: 0 == 1 is true?

Rails
1

I must be missing something fundamental. In the test code below, the
assertion should fail unless WebPageCache.count == 1. But
WebPageCache.count is zero (according to the print statement and all
other evidence), yet the assertion passes.

What am I missing?

class WebPageCacheTest < ActiveSupport::TestCase
…
test “table has one entry” do
p “<<< #{WebPageCache.count}” # prints ‘<<< 0’
assert (WebPageCache.count == 1), “WebPageCache size is not 1”
end
…
end

(An aside: if I can’t trust a simple assert, what CAN I trust? :slight_smile:

  • ff
2

On Fri, May 7, 2010 at 4:39 PM, Fearless F. [email protected]
wrote:

I must be missing something fundamental. In the test code below, the
assertion should fail unless WebPageCache.count == 1.

assert (WebPageCache.count == 1), “WebPageCache size is not 1”

Did you try WebPageCache.count.to_i == 1 or WebPageCache.count == ‘1’ ?

–
Greg D.
destiney.com | gregdonald.com

3

Greg D. wrote:

Did you try WebPageCache.count.to_i == 1 or WebPageCache.count == ‘1’ ?

For grins, yes. No change.

But I’m beginning to guess: are the db tables rolled back between EVERY
call to test? I thought they were rolled back at the end of the test
run, not between every test.

Asked another way: if web_page_caches starts as empty, what are the
expected results in the following?

======
class WebPageCacheTest < ActiveSupport::TestCase
test “store an entry” do
WebPageCache.store(“my_key”, “my_value”)
assert (WebPageCache.count == 1), “WebPageCache size is not 1”

test “table still has one entry” do
assert (WebPageCache.count == 1), “WebPageCache size is not 1”
end
end

4

On May 7, 11:09 pm, Fearless F. [email protected] wrote:

Greg D. wrote:

Did you try WebPageCache.count.to_i == 1 or WebPageCache.count == ‘1’ ?

For grins, yes. No change.

But I’m beginning to guess: are the db tables rolled back between EVERY
call to test? I thought they were rolled back at the end of the test
run, not between every test.

between every test (or rather each test runs inside a transaction
which is rolled back). There should be no leakage of state from one
test to another.

Fred