Salvador Fandiño wrote: > > And BTW I think that 'use assertions' without an import list should > > be disallowed. > > if no keys are passed the current package name is used as the filter key. Oops, you're right. Nevermind. > Other unresolved issue is what to do with assertions appearing before > any "use assertions" declaration, maybe to print a warning? Yes, that would be a good idea : something like "assertion never activated at ..." > > I have one concern : > > if you say "use assertions qw/foo bar/" you need to pass -Afoo,bar > > to activate the assertions in the corresponding block. I'd have > > expected that -Afoo and -Abar work too (i.e. that assertion groups > > are or-ed, not and-ed). > > well, I have found that it's more useful to have 'and' selection than > 'or' selection as this allows to mantain several orthogonal criterions. > i.e., if you have an application with different layers you can use the > layer names as one criterion and object life-cicle can be other > criterium (contructors, destructors, storing, etc). When you debug the > app. it is possible you would like to see the destructors for layer III, > but not all the destructors calls and all the layer III calls at the > same time. Yes but if you've said "use assertions qw/destructors layerIII/" and then -Adestructors, your assertions won't be asserted. No ? I just feel that the interface is not very flexible. On the other hand it's possible to say -A'destructors|layerIII' to activate both. Opinions ?Thread Previous | Thread Next