On 12/18/05, demerphq <[EMAIL PROTECTED]> wrote: > On 12/17/05, chromatic <[EMAIL PROTECTED]> wrote: > > On Saturday 17 December 2005 08:23, demerphq wrote: > > > > > It seemed to me that > > > a better patch would be to change the way harness handles directives > > > so it recognizes TODO & SKIP as being a valid directive. > > > > What would that mean? SKIP tests don't run. TODO tests do. > > > > If the test doesn't run, I think it's a SKIP and nothing else. > > Well, Test::Harness aka the opposite point of view and considers it a > TODO and not a SKIP at all as the TODO is first. > > And Test::More::todo_skip() outputs "not ok" for the test, but > test.pl::todo_skip() emits "ok" for the test, which Test::Harness > treats as an unexpected success. If Test::Harness realized that TODO & > SKIP meant both then it could differentiate between the case of > todo_skip() and the case of a real todo that unexpectedly succeeds or > fails. After all it needs to support scenarios where the output isnt > being manufactured by either Test::Builder or test.pl, so you can't > just rely on fixing either. (IMO anyway) > > BTW, none of this is my logic, its legacy code. If you want to > question whether it makes sense to have something be both TODO and > SKIP at the same time ask the person that wrote todo_skip() in the > first place. > > :-)
Anyway, i hope all of these patches aren't warnocked because of the Test::Harness discussion. The facts are that test.pl::todo_skip() and Test::Harness don't play together properly and thus T::H reports things as unexpectedly passed when they aren't. Yves -- perl -Mre=debug -e "/just|another|perl|hacker/"
