Bernhard Schmalhofer wrote:
I added this failing, as the test '1 equals 1' in
languages/lisp/t/arithmetics.t started failing after
the merge. Investigating that failure, I found that I could reproduce
the problem with a simple
PIR test case. This is what I added in t/pmc/objects.t. So it's a test
for something that was previously
working.
All of that is incredibly valuable information, which helped me debug
the problem once I had it.
Yes, that's the intention of 'TODO' as I understood. Tests that are
failing because something isn't done
yet. But IMHO somebody should look at the issue and classify it as TODO.
IMHO the classifier doesn't have to be submitter of the test case, as
there should be
a low entrance barrier for adding tests.
As a general rule of thumb, if you find a failure but don't have the
time/expertise to debug it or classify it as TODO or BUG or "new
underdocumented feature", but do have enough time to create a failing
test for the problem (always nice), submit a ticket explaining the
problem and attach the test to the ticket.
The patch monster/cage cleaner who handles the ticket can decide if the
test is good and where/whether it should be added to the test files.
Allison
