redbull2012 has asked for the wisdom of the Perl Monks concerning the following question:

Hello Monks,

Below are some snippet codes that i don't understand the behavior, please help:

First case:

my $foo = "weird123"; my $result = $foo.($foo =~ s/123//?$foo:""); print $result; # "weirdweird"

Second case:

my $foo = "weird123"; my $result = "".$foo.($foo =~ s/123//?$foo:""); #or: my $result = $foo."".($foo =~ s/123//?$foo:""); print $result; # "weird123weird"

I already had a discussion with one fellow monk, he thought that the "" is forcing the left side of "." executed first before the substitution. He was also not sure so i ask this to make us understand better.

Thanks in advance!

Replies are listed 'Best First'.
Re: Strange behavior
by ikegami (Patriarch) on Jul 03, 2014 at 04:43 UTC

    Your guess is incorrect. It's undocumented and potentially undefined, but the concatenation operator always evaluates its LHS before its RHS.

    The key is that the variable is placed on the stack, not its value.


    First snippet (simplified):

    $ perl -MO=Concise,-exec -E'my $foo = "weird123"; say $foo . do { $foo + =~ s/123//; $foo };' ... 6 <;> nextstate(main 49 -e:1) v:%,{,469764096 7 <0> pushmark s 8 <0> padsv[$foo:47,49] s 9 <0> enter s a <;> nextstate(main 48 -e:1) v:%,{,469764096 b <0> padsv[$foo:47,49] sRM c <$> const[PV ""] s d </> subst(/"123"/) vKS e <;> nextstate(main 48 -e:1) v:%,{,469764096 f <0> padsv[$foo:47,49] s g <@> leave sKP h <2> concat[t2] sK/2 i <@> say vK ...
    1. (8) $foo is placed on the stack. (The scalar, not its value.)
    2. (d) $foo is changed.
    3. (f) $foo is placed on the stack. (The scalar, not its value.)
    4. (h) The values on the on the stack ($foo and $foo) are replaced with the result of their concatenation.

    Replicating the behaviour with an inspectable stack:

    use feature qw( say ); use Data::Alias qw( alias ); sub concat { my $lhs = shift; my $rhs = shift; $lhs . $rhs } { my $foo = "weird123"; alias push @_, $foo; $foo =~ s/123//; alias push @_, $foo; alias push @_, &concat; say shift(@_); }

    Second snippet (simplified):

    $ perl -MO=Concise,-exec -E'my $foo = "weird123"; say $foo . "" . do { + $foo =~ s/123//; $foo };' ... 6 <;> nextstate(main 49 -e:1) v:%,{,469764096 7 <0> pushmark s 8 <0> padsv[$foo:47,49] s 9 <$> const[PV ""] s a <2> concat[t2] sK/2 b <0> enter s c <;> nextstate(main 48 -e:1) v:%,{,469764096 d <0> padsv[$foo:47,49] sRM e <$> const[PV ""] s f </> subst(/"123"/) vKS g <;> nextstate(main 48 -e:1) v:%,{,469764096 h <0> padsv[$foo:47,49] s i <@> leave sKP j <2> concat[t3] sKS/2 k <@> say vK ...
    1. (8) $foo is placed on the stack. (The scalar, not its value.)
    2. (9) An empty string is placed on the stack.
    3. (a) The values on the on the stack ($foo and the new value) are replaced with the result of their concatenation.
    4. (f) $foo is changed.
    5. (h) $foo is placed on the stack. (The scalar, not its value.)
    6. (j) The values on the on the stack (the result of the earlier concatenation and $foo) are replaced with the result of their concatenation.

    Replicating the behaviour with an inspectable stack:

    use feature qw( say ); use Data::Alias qw( alias ); sub concat { my $lhs = shift; my $rhs = shift; $lhs . $rhs } { my $foo = "weird123"; alias push @_, $foo; alias push @_, ""; alias push @_, &concat; $foo =~ s/123//; alias push @_, $foo; alias push @_, &concat; say shift(@_); }

      Thanks for clarifying that! It was my guess that the "" was forcing $foo to be interpolated into the string before the substitution operation was performed on it.

      You must always remember that the primary goal is to drain the swamp even when you are hip-deep in alligators.

        What interpolation???

        In short:

        In the first snippet, you are concatenating $foo with $foo, so that's what you get.

        In the second snippet, you are concatenating something derived from $foo and $foo, so that's what you get.

      Thank to your detail explanation, I have learnt a lot from this.