In article <[EMAIL PROTECTED]>, Gabor
Szabo <[EMAIL PROTECTED]> wrote:
> So first I thought to try to run test.pl before the t/-tests. Hence the
> subject. How to do this is still interesting, though maybe not necessary:
You used to be able to set the order of tests through MakeMaker, then
that feature disappeared. I wrote Test::Manifest to get that back. I
specify which tests and in which order I want to run them in the
t/test_manifest file.
In the t/test_manifest file, I can add ../test.pl as the first test
to run.
> Later I played a bit with the t-tests to see what happens if they fail.
> I saw that all of the cases I checked when the t/* test failed the test.pl
> suit did not run at all.
If you run them all as the same command from the Makefile (rather than
one for t tests and one for test.pl in the test_dynamic target), make(1)
doesn't get the chance to abort when it encounters an error. In your
case, the situation is the same as this snippet, where the echo does
not get a chance to run because make(1) stops at the "error".
all:
false
echo "just another Makefile hacker,"
> Meaning that I got the error message right in my
> face. This is a good thing. My questions: can I rely on this ?
i don't know all the different versions of make(1) and their behavior,
so I can't answer that one. It has never been a problem for me,
though,
--
brian d foy, [EMAIL PROTECTED]
