[m-rev.] for review: delete obsolete procedures from string module

Julien Fischer jfischer at opturion.com
Thu Apr 14 10:43:13 AEST 2022


On Wed, 13 Apr 2022, Zoltan Somogyi wrote:

> 2022-04-13 22:57 GMT+10:00 "Julien Fischer" <jfischer at opturion.com>:
>> There are tests in that directory that emit warnings,
>
> A grep for "Warning" in tests/invalid*/*err_exp yields 33 matching files.
> I also don't know whether they are intentional.

I have had a quick look, the warnings emitted by tests in that directory are one of:

1. Unavoidable due to the nature of the test case.

2. Unrelated to the what is being tested.  This is mostly a result of
new warnings having been added to the compiler since he test was added.

3. Are the *only* output of test, e.g. foreign_singleton.m,
empty_interface.m.

>> I haven't looked at whether those warnings are germane to what is being
>> tested.
>
> In this case, it is germane; the existing test tests the handling of mode-specific
> clauses.
>
>> However, my feeling is that specific tests for warnings ought
>> to go into tests/warnings.  I would be inclined to disable the new
>> warning for the test in tests/invalid
>
> Traditionally, we have not done that in situations like this, as the 33 hits
> above shows.

I don't think that's the result of any particular policy on our part
though.  I suggest the policy should be:

     tests of errors go in the invalid* directories
     tests of warnings go in the warnings directory
     extraneous output should be suppressed as far as possible.
     exceptions to any of this should be documented in the test case

I will look into cleaning up at least (2) and (3) above.

Julien.


More information about the reviews mailing list