----- Original Message ---- > From: Andy Lester <[EMAIL PROTECTED]>
> I guess I'm not seeing why a deferred plan is better than no plan at > all. Seems to me the whole point of a plan is that you know up front > how many they're gonna be. I've not explained myself well. Sorry about that. The reason you have to know up front is because that's how the entire system was designed. We currently have two cases when we should have at least three: 1. We know the test count up front, in which case we declare a leading plan. 2. We don't know the test count up front, in which case Test::Builder supplies a trailing plan which merely tells me how many tests I've run. This misses the obvious case where I don't know the expected count up front, but I do know the expected count at the end (not guaranteed to be the same as the actual count). It should be trivial to fix Test::Builder and co., so that the programmer supplies the trailing plan instead of Test::Builder. That being said, I like Adrian's code for this and I'll be stealing it. Cheers, Ovid -- Buy the book - http://www.oreilly.com/catalog/perlhks/ Perl and CGI - http://users.easystreet.com/ovid/cgi_course/ Personal blog - http://publius-ovidius.livejournal.com/ Tech blog - http://use.perl.org/~Ovid/journal/
