On 05/07/2013 11:42 PM, Eric Blake wrote:
The surest way to avoid regressions is to test documented behavior :)
* tests/virstringtest.c (testStrdup): New test case.
Signed-off-by: Eric Blake <eblake@redhat.com> --- tests/virstringtest.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 48 insertions(+), 1 deletion(-)
...
+ value = VIR_STRNDUP(array[i++], NULL, 5); + if (value != 0) { + fprintf(stderr, "unexpected strdup result %d, expected 0\n", value); + goto cleanup; + } + if (i != 2) { + fprintf(stderr, "unexpected side effects i=%zu, expected 2\n", i); + goto cleanup; + }
Should we test the side effects for the other two arguments as well?
+ if (STRNEQ_NULLABLE(array[0], "hello") || array[1]) { + fprintf(stderr, "incorrect array contents '%s' '%s'\n", + NULLSTR(array[0]), NULLSTR(array[1])); + goto cleanup; + } + + ret = 0;
ACK Jan