Test::unit assertion pass scenario

[John Smith]

When using the test::unit assertion, such as assert_equal, the script
will throw a failure if the test condition (assertion) is not met. If it
does pass, no output is displayed.

Is there a way to force the results of the test to display both passes
and failures?

Thanks in advance!

 

[Ryan Davis]

When using the test::unit assertion, such as assert_equal, the script
will throw a failure if the test condition (assertion) is not met. If = it
does pass, no output is displayed.
=20
Is there a way to force the results of the test to display both passes
and failures?

how are you using it?? Normally it displays something like:

/System/Library/Frameworks/Ruby.framework/Versions/1.8/usr/bin/ruby -w =

-I../../minitest/dev/lib:lib:ext:bin:test -e 'require "rubygems"; =
require "minitest/autorun"; require "test/test_autotest.rb"; require =
"test/test_focus.rb"; require "test/test_unit_diff.rb"; require =
"test/test_zentest.rb"; require "test/test_zentest_mapping.rb"'=20

Loaded suite -e
Started
= ..........................................................................=
.............................
Finished in 0.214105 seconds.
=20
103 tests, 259 assertions, 0 failures, 0 errors, 0 skips

(this is for minitest, not test/unit, but the output is very similar)

Your test should be set up like:

# test_blah.rb:
require 'test/unit'
=20
class TestThingy < Test::Unit::TestCase
def test_thingy
assert_equal 2, 1+1
end
end.

Here is a run:

 

[John Smith]

Yep, the example below is exactly the way I'm using it.
However, as demonstrated in your example, the 259 assertions that were
run (and passed) do not display any kind of passing checkpoint, the way
it would have if any of those assertions failed.

Basically, I am looking for a way to provide info just for both passed
and failed assertions, similar to what is done when an assertion fails.

Thanks again!

 

[Ryan Davis]

Yep, the example below is exactly the way I'm using it.
However, as demonstrated in your example, the 259 assertions that were=20=

run (and passed) do not display any kind of passing checkpoint, the = way=20
it would have if any of those assertions failed.
=20
Basically, I am looking for a way to provide info just for both passed=20=

and failed assertions, similar to what is done when an assertion =

fails.

Your use of "info" is pretty nebulous.

..........................................................................=
.............................

All of that is "info".

What do you want it to do differently, and (more importantly) WHY?

 

[Phillip Gawlowski]

Yep, the example below is exactly the way I'm using it.
However, as demonstrated in your example, the 259 assertions that were
run (and passed) do not display any kind of passing checkpoint, the way
it would have if any of those assertions failed.

Basically, I am looking for a way to provide info just for both passed
and failed assertions, similar to what is done when an assertion fails.

Why?

If an assertion passes everything is well (within the test parameters,
anyway ), and no action is required.

In fact, you'd drown out, since you degrade the signal/noise ratio,
actual failures.

If you are looking for a way to see what code gets exercised (and if all
code gets tested), RCov (used to be) a good solution (alas, it hasn't
been updated since 2007) to test for code coverage, too.

 

[Ryan Davis]

If you are looking for a way to see what code gets exercised (and if =

all code gets tested), RCov (used to be) a good solution (alas, it =
hasn't been updated since 2007) to test for code coverage, too.

not true:

rcov (0.9.7.1)
Platform: ruby, java
Authors: Relevance, Chad Humphries (spicycode), Aaron Bedra
(abedra), Jay McGaffigan, Mauricio Fernandez
Homepage: http://github.com/relevance/rcov
=20
Code coverage analysis tool for Ruby

see http://gemcutter.org/gems/rcov

Versions
=95 0.9.7.1 December 29, 2009
=95 0.9.7.1 December 29, 2009 java
=95 0.9.7 December 27, 2009
=95 0.9.7 December 27, 2009 java
=95 0.9.6 May 12, 2009

But I feel I should point out: rcov doesn't tell you that your tests are =
any good... it is only good for "what code gets exercised" but not =
"[all] code gets tested".

 

[John Smith]

Why is a good question. First, the extra info is not for myself, nor
would it be for any of the devs who may run it. The theory is that
anyone who writes the test or uses them regularly should be familiar
with what is being tested anyway, and hence, only the failures really
need further investigation.

It's more of a CYA item for those who are, shall we say, not in the
know.

 

[Jörg W Mittag]

But I feel I should point out: rcov doesn't tell you that your tests
are any good... it is only good for "what code gets exercised" but
not "[all] code gets tested".

Simple proof: take a hypothetical "perfect" test suite with 100%
coverage. Remove all assertions. Still 100% coverage, but *nothing*
gets tested anymore.

jwm

 

[Ryan Davis]

But I feel I should point out: rcov doesn't tell you that your tests
are any good... it is only good for "what code gets exercised" but
not "[all] code gets tested".

=20
Simple proof: take a hypothetical "perfect" test suite with 100%
coverage. Remove all assertions. Still 100% coverage, but *nothing*
gets tested anymore.

_Exactly_

 

[Ryan Davis]

Why is a good question. First, the extra info is not for myself, nor=20=

would it be for any of the devs who may run it. The theory is that=20
anyone who writes the test or uses them regularly should be familiar=20=

with what is being tested anyway, and hence, only the failures really=20=

need further investigation.
=20
It's more of a CYA item for those who are, shall we say, not in the=20
know.

Some sort of detailed report of exactly what assertions you're running =
isn't a very good CYA. You might be better off with:

+ # of tests
+ # of assertions (or better: assertions / test)
+ % of coverage (possibly add heckle #'s, but that's a serious PITA)
+ loc test / loc impl (but please for gods' sake refactor both sides)
+ test time

and then graph that over time.

 

[Phillip Gawlowski]

not true:

Someone needs to update eigenclass.org's RCov page, then.

 

[Ryan Davis]

all code gets tested), RCov (used to be) a good solution (alas, it =
hasn't been updated since 2007) to test for code coverage, too.

=20
Someone needs to update eigenclass.org's RCov page, then.

He's not responding to anyone. Which is why it has new parents.

 

[Phillip Gawlowski]

He's not responding to anyone. Which is why it has new parents.

Aha! Thanks.