This is because skip_all exists immediately but Test::NoWarnings has installed an END block that will run a test anyway. I suppose I should change T::NW to detect "skip_all" and not run its test but at the moment there is no T::B API for detecting "skip_all".
MScwern, do you want to add one? What about BAIL_OUT? F 2008/5/15 Gabor Szabo <[EMAIL PROTECTED]>: > Yesterday I was happy to notice that you can use Test::Warn together with > Test::NoWarnings. > So you can test for a watning in a specific test while testing that > nothing else gives a warning.\ > Good. :-) > > Today I found out that if you skip_all tests while you have > Test::NoWarnings your test > will fail. > Bad. :-( > > > Gabor > > > The following code: > > =============== > > use Test::More; > use Test::NoWarnings; > > plan skip_all => 'Why not?'; > plan tests => 2; > > ok(1) > > =============== > > gives this output: > > 1..0 # Skip Why not? > You tried to run a test without a plan at > /root/perl5lib/lib/Test/NoWarnings.pm line 45. > END failed--call queue aborted. > > =============== > > and prove gives this: > > =============== > > t/w......You tried to run a test without a plan at > /root/perl5lib/lib/Test/NoWarnings.pm line 45. > END failed--call queue aborted. > t/w......skipped: Why not? > > Test Summary Report > ------------------- > t/w.t (Wstat: 65280 Tests: 0 Failed: 0) > Non-zero exit status: 255 > Files=1, Tests=0, 0 wallclock secs ( 0.01 usr 0.01 sys + 0.04 cusr > 0.00 csys = 0.06 CPU) > Result: FAIL > > > =============== >
