Closed GoogleCodeExporter closed 8 years ago
That's not good. Let's see what we can do for v3.0.6 Update 2.
Original comment by jeff.br...@gmail.com
on 27 Apr 2009 at 6:08
Ok, so in general the problem is that if fixture setup or teardown fails then
the
children in the test tree just don't ever get a chance to run. This is a
problem not
just for assembly fixtures and test fixture but other things too, like suites.
I guess the solution will either be to generate fake failure results for tests
that
did not run or to promote the fixture to "test case" status so it shows as a
failure
in the stats.
The problem is that the statistics will still be wrong when any data-driven
tests
were not run. For example, if all N runs of a data-driven test (eg. row-test)
failed
you would see N failures logged. However if an assembly fixture fails
initialization
then in general we cannot enumerate the whole list of runs there would have
been so
we don't know how many failures to report...
Consequently, I'm thinking I might need to introduce a new TestStatus like
"blocked"
or "fatal". The idea here is that when we cannot initialize something like an
assembly fixture, we just give up on that whole part of the test tree and
report the
meta-failure loudly.
In any case, fixing this is too risky for v3.0.6 so I'll defer it in v3.0.7
instead.
Original comment by jeff.br...@gmail.com
on 28 Apr 2009 at 3:31
Note that marking a fixture [Ignored] also has similar effects: we don't get
separate
ignored results in the stats for each test in the fixture...
Original comment by jeff.br...@gmail.com
on 28 Apr 2009 at 3:33
This sounds like a good way to fix it; I wasn't particularly concerned about
the numbers being correct; it is more
that it doesn't report any failure at all - which is very problematic for
continuous integration as it doesn't detect
any problem and we don't get alerted, and it treats it as a success; when it
could be (and likely is) a serious issue
if the AssemblyFixture cannot be run.
Original comment by mab...@gmail.com
on 30 Apr 2009 at 9:06
Original comment by jeff.br...@gmail.com
on 29 Jun 2009 at 8:33
Let's see what we can do.
Original comment by jeff.br...@gmail.com
on 31 Jul 2009 at 10:44
The fix now is to promote a fixture to the status of "test case" if it fails and
appears to have prevented its child tests from failing.
Bottom line, you will see a failure in the report statistics now.
Original comment by jeff.br...@gmail.com
on 14 Aug 2009 at 11:32
Original issue reported on code.google.com by
mab...@gmail.com
on 14 Apr 2009 at 10:55