Re^3: BEGIN and compile-time

It is more appropriate to test $@.

eval "use Foo::Bar";
if ($@) {
    warn "Eval failed: $@";
}
[download]

Comment on Re^3: BEGIN and compile-time Download Code
Re^4: BEGIN and compile-time ($@) by tye (Cardinal) on Nov 10, 2006 at 19:39 UTC
It is more appropriate to test $@. Oh, I disagree. There are too many cases where $@ doesn't survive. It is more fool-proof (and one of my "best practices") to check the return value of `eval "....; 1"` to determine whether or not eval failed (and then to use $@ to report what the error was, though $@ might no long contain this information). - tye	[reply] [d/l]
Re^5: BEGIN and compile-time ($@) by imp (Priest) on Nov 10, 2006 at 19:57 UTC
That's a good point. Could you provide an example of where a failed `eval "use Foo"` does not result in $@ being populated?	[reply] [d/l]
Re^6: BEGIN and compile-time ($@) by tye (Cardinal) on Nov 10, 2006 at 20:01 UTC
It usually involves a "broken" object DESTROY method, usually one that uses `eval` but forgets to `local( $@ )`. - tye	[reply] [d/l]
Re^4: BEGIN and compile-time by ikegami (Pope) on Nov 10, 2006 at 19:40 UTC
I like `eval { require Foo::Bar } or warn "Eval failed: $@";` [download] Since `require` only returns true or an exception. The `import` can be done later, although I usually don't want to import from a dynamically loaded module. Update: Oops, silly me! Replaced `eval ""` with `eval {}`. Thanks tye.	[reply] [d/l] [select]


Clear questions and runnable code get the best and fastest answer
	PerlMonks