Until now, teardown stdout/stderr output was not reported upon test failure.
However such output is sometime necessary to understand the failure.
fix#442
pytest.skip() must not be used at module level because it can easily be
misunderstood and used as a decorator instead of pytest.mark.skip, causing the
whole module to be skipped instead of just the test being decorated.
This is unexpected for users used to the @unittest.skip decorator and therefore
it is best to bail out with a clean error when it happens.
The pytest equivalent of @unittest.skip is @pytest.mark.skip .
Adapt existing tests that were actually relying on this behaviour and add a
test that explicitly test that collection fails.
fix#607
Add --continue-on-collection-errors option to restore the previous behaviour:
Execute tests (that were successfully collected) even when collection errors
happen.
Some tests had to be modified e.g. because the return code changed to 2
(EXIT_INTERRUPTED) instead of 1 (EXIT_TESTSFAILED) because an Interrupted
exception is raised on collection error.
Implemented via pair programming with:
Oleg Pidsadnyi <oleg.pidsadnyi@gmail.com>
closes#1421
This appears to have been unused since commit
320835d "split out pytest-xdist related reporting to the plugin"
in July 2010. It's the only caller outside of _pytest/runner.py
of the `call_and_report` helper function there, so cutting it out
makes that more of a pure helper function and makes it slightly
easier to understand the code in _pytest/runner.py .
Passing tests override that default, making the color green; but several other
"boring" statuses (xfailed, xpassed, deselected, skipped) have no effect.
Net effect: if only "boring" tests are seen, or no tests at all, the summary
bar is yellow.
refactoring how nodeid's are constructed. They now are always
relative to the "common rootdir" of a test run which is determined by
finding a common ancestor of all testrun arguments.
--HG--
branch : issue616
This strips the line number, /@\d/, from the verbose output so it is
directly the node ID of the test. This in turn means no special logic
for accepting the line number as part of the node ID is needed when
parsing the command line.
The new "--tb=auto" option (default) will only display long tracebacks
for the first and last entry. You can get the old behaviour of printing
all entries as long entries with "--tb=long". Also short entries by
default are now printed very similarly to "--tb=native" ones.
Bruno Oliveira
fbjorn
Michal Wajszczuk
Mathieu Clabaut
Daniel Hahler
Ronny Pfannschmidt
RedBeardCode
aostr
Omar Kohl
Greg Price
Ronny Pfannschmidt
David Vierra
Florian Bruhin
Florian Bruhin
Eric Siegerman
holger krekel
Floris Bruynooghe