Re: sc_tests_list_consistency failure when adding tests [was: df --output]

[Jim Meyering]

Bernhard Voelker wrote:
> On 09/21/2012 11:28 AM, Jim Meyering wrote:
>> Bernhard Voelker wrote:
>>> I wonder if syntax-check couldn't prevent this ... didn't we have this
>>> already a few weeks ago?
>>
>> Good idea.
>> It's definitely worth a try.  It will depend on how many
>> false positives there are.
>
> Apropos: the syntax-check failed yesterday, when I tried to
> add the new test, because the test was not yet in the git history,
> but sc_tests_list_consistency seemed to check on that.
>
> So I committed despite of that error, and a subsequent syntax-check
> passed. Is this normal behavior? I didn't experience such before the
> non-recursive work. And this is the first test to be added
> since then ... (but maybe it was just too late at night for me).

For reference, here's the rule:

# Ensure that all version-controlled test cases are listed in $(all_tests).
sc_tests_list_consistency:
        @bs="\\";                                                       \
        test_extensions_rx=`echo $(TEST_EXTENSIONS)                     \
          | sed -e "s/ /|/g" -e "s/$$bs./$$bs$$bs./g"`;                 \
        {                                                               \
          for t in $(all_tests); do echo $$t; done;                     \
          cd $(top_srcdir);                                             \
          $(SHELL) build-aux/vc-list-files tests                        \
            | grep -v '^tests/init\.sh$$'                               \
            | $(EGREP) "$$test_extensions_rx\$$";                       \
        } | sort | uniq -u | grep . && exit 1; :


If you merely "git add" your new test script, that should be enough,
since that test compares the list of VC'd files under tests/
and the list of test file names.

In other words, afaik, there is no problem.