Bug#703250: apt: failing tests in test/libapt does not cause FTBFS
On 18 March 2013 15:25, Michael Vogt <mvo@debian.org> wrote:
>> This makes it so the first failure prevents later tests in the same
>> file from running. It can be bothersome to diagnose faults if e.g.
>> individual tests are not properly ordered, or are otherwise
>> independent of each other. Some test frameworks avoid such early
>> abort and run all tests.
>
> Great! I like this better than my original approach of failing too
> early. Attached is a small variation that will also count the number
> of failures, the obvious issue is that failure numbers > 255 in a
> single test can not be reported via the exit code. The diff is against
> your patch (and also against the "debian-sid" branch in bzr).
FWIW I would use the second patch I sent, as it uses similar
terminology and semantics as Niels and is thus consistent between the
different test suites.
Using exit status to report number of failures is generally
discouraged, no? C.f. standards.info and common practice.
Reply to: