prototypes parens/parenthesis hash initialization/declaration / Too many arguments for / $$

[Anonymous Monk]

I'm confused, I thought this would work

$ perl -MData::Dump -e " sub ff($$){ qq{@_} } my %f = ( 12 => ff 1,2, 
+34 => ff 3,4 ); dd\%f "
Too many arguments for main::ff at -e line 1, near "4 )"
Execution of -e aborted due to compilation errors.

without having to type parens like

$ perl -MData::Dump -e " sub ff($$){ qq{@_} } my %f = ( 12 => ff( 1,2)
+ , 34 => ff( 3,4) ); dd\%f "
{ 12 => "1 2", 34 => "3 4" }

A prototype of ($) doesn't require parens ... anyone?

[choroba]

Another reason why people should not use prototypes. They do not do what you think. See Are prototypes evil?.

Update: BTW, the mentioned behaviour is normal:

$ perl -e 'print atan2 1, 2, 3'
Too many arguments for atan2 at -e line 1, at EOF

[Anonymous Monk]

Another reason why people should not use prototypes. They do not do what you think. See Are prototypes evil?

Oh, context coersion gone amuch, icky

I guess if that worked there wouldn't be a need for Devel::Declare and the like

[AnomalousMonk]

Why not try a different approach?

>perl -wMstrict -MData::Dump -le
"sub ff {
   return
     map { $_ => join ' ', m{ \d }xmsg }
     map { m{ \A \d{2} \z }xms ? $_ : die qq{bad arg: '$_'} }
     @_;
   }
 ;;
 my %f = ff(12, 34);
 dd \%f;
"
{ 12 => "1 2", 34 => "3 4" }

[kcott]

"I'm confused, I thought this would work ..."

Using -MO=Deparse,-p will show you how Perl parses the code and why it doesn't work:

$ perl -MO=Deparse,-p -e 'sub ff($$){ qq{@_} } my %f = ( 12 => ff 1,2,
+ 34 => ff 3,4 );'
Too many arguments for main::ff at -e line 1, near "4 )"
-e had compilation errors.
sub ff ($$) {
    "@_";
}
(my(%f) = (12, &ff(1, 2, 34, ff(3, 4))));

On the last line of the output, you'll see &ff(1, 2, 34, ff(3, 4)) has 4 arguments - your prototype specifies 2.