chore: output unambiguous message

[hyoklee]

See also HDFGroup/vol-daos#59

[jhendersonHDF]

This isn't testing H5Pclose in particular. These messages refer to testing that resources can be cleaned up after a multi-part test so that the user knows whether testing failed during the test, or during cleanup of resources.

[hyoklee]

What's your definition of cleanup? Who wrote this code?

/*                                                                                
 * Cleanup temporary test files                                                   
 */
static void
cleanup_files(void)
{
    hid_t fapl_id = H5I_INVALID_HID;

    if ((fapl_id = create_mpi_fapl(MPI_COMM_WORLD, MPI_INFO_NULL, TRUE)) < 0) {
        if (MAINPROCESS)
            HDprintf("    failed to create FAPL for deleting test files\n");
        return;
    }

    H5Fdelete(FILE_CREATE_TEST_FILENAME, fapl_id);

[jhendersonHDF]

I wrote that code and it's specifically for cleaning up files generated in the parallel file tests, which are separate from the vol_test_parallel.h5 file that contains all the other tests like the dataset ones. Cleanup of resources is cleanup.

[hyoklee]

What do you think the meaning of "cleanup" here from 1998?

/*-------------------------------------------------------------------------
 * Function:  h5_cleanup
 *
 * Purpose:  Cleanup temporary test files.
 *    base_name contains the list of test file names.
 *    The file access property list is also closed.
 *
 * Return:  Non-zero if cleanup actions were performed; zero otherwise.
 *
 * Programmer:  Albert Cheng
 *              May 28, 1998
 *
 *-------------------------------------------------------------------------
 */
int
h5_cleanup(const char *base_name[], hid_t fapl)
{
    int retval = 0;

    if (GetTestCleanup()) {
        /* Clean up files in base_name, and the FAPL */
        h5_clean_files(base_name, fapl);

        retval = 1;
    } /* end if */

    /* Restore the original error reporting routine */
    h5_restore_err();

    return retval;
} /* end h5_cleanup() */

[jhendersonHDF]

Nothing in "cleanup" refers specifically to files.

[hyoklee]

Then, why do you say clean up here?

On the other hand, the parallel test does seem to be missing a call to H5Fdelete to clean up after itself

[jhendersonHDF]

Because in that particular instance I was referring specifically files.