Re^2: Safer monkey-patching

[tobyink]

Redefining a function only produces a warning if "use warnings" is in effect. So classic monkey patching will not necessarily produce compile-time warnings.

If a conflict warning is desired, that's fairly easy to add in with my method...

BEGIN {
   require Example::Error;
   foreach my $method (qw/asplode/) {
     warn "Example::Error->$method already defined."
       if Example::Error->can($method);
   }
   push @Example::Error::ISA, __PACKAGE__;
 }

[JavaFan]

Redefining a function only produces a warning if "use warnings" is in effect. So classic monkey patching will not necessarily produce compile-time warnings.

Are you serious? You consider that a reasonable argument? "You shouldn't monkey patch because you may not get a warning on a name clash if you don't enable warnings"?

I'm pretty sure that anyone who knows how to monkey patch knows about warnings.

If a conflict warning is desired, that's fairly easy to add in with my method...

Goody. Additional scaffolding, and you still aren't any further than what can be achieved with monkey-patching.

Assuming Example::Error doesn't use AUTOLOAD to implement asplode (heh, if you want to consider monkey patchers that don't enable warnings, I will consider AUTOLOAD), it still doesn't solve anything. Once Example::Error implements asplode, you get a warning, and where your code expects to call your asplode, it calls Example::Error::asplode. Ergo, you haven't solved anything. It isn't safer than monkey-patching. With the extra scaffolding, it isn't unsafer either.