In the case of nested maps, keeping track of what each $_ is holding involves an extra cognitive load. It's simply that it takes the reader a fair bit longer than normal to figure out the intent, and for that reason, I'd prefer to avoid it.

I agree with that. It would be nice if map (and grep) could take a variable like foreach does.

# this doesn't work
my @joined
    = map my $url { map my $code { $url . $code } @codes } @urls;
[download]

I guess, but I actually find that version harder to read than the first version. For one thing, it has no indentation options that aren't kind of terrible.

For the general, non-nested versions, it might be nicer.

for(split(" ","tsuJ rehtonA lreP rekcaH")){print reverse . " "}print "\b.\n";

If the map BLOCK is more complicated than [insert personal threshold here], or they are nested I just declare the variable myself.

@joined = map {
  my $url = $_;
  map {
    my $code = $_;
    $url . $code;
  } @codes
} @urls;
[download]

...nothing syntactically wrong with a sentence like "The cat the dog the man hit bit died."

Ah. Fluent reverse Polish. Or is it Yoda egg-nog on (too much) ?

I've tried to parse this... putting the brackets back in I get: died(the cat bit (the dog hit the man)) -- I'm assuming: (a) that each 'the' is part of the name of each object/subject; (b) that 'died' is a unary (intransitive); and (c) that 'bit' and 'hit' are binary operations (transitive).

So much for syntax... semantics-wise, assuming that the 'hit' and 'bit' operators return the object modified by the operation inflicted upon it by the subject...

...no wonder the man died. What on earth had he done to upset both the cat and the dog ? And what did the dog hit him with ?

Of course, being reverse polish the operators may themselves be reverse -- so the cat gets it because the man hit the dog ? What kind of twisted world is this ?

"The cat the dog the man hit bit died."

.oO(Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo.)

And you didn't even know bears could type.

Perl6 provides a pretty handy way to do this. It has X which I beleive is pronounced "cross" so that you can do: (source)

use v6;

my @urls = ('http://www.something.com/blah.aspx?code=',
            'http://www.somethingelse.com/stuff.aspx?thing=');

my @ids = ('375035304','564564774','346464646');

my @combined = (@urls X @ids).map: {$^a ~ $^b};

.say for @combined;
[download]

Which outputs the following on Rakudo already:

http://www.something.com/blah.aspx?code=375035304
http://www.something.com/blah.aspx?code=564564774
http://www.something.com/blah.aspx?code=346464646
http://www.somethingelse.com/stuff.aspx?thing=375035304
http://www.somethingelse.com/stuff.aspx?thing=564564774
http://www.somethingelse.com/stuff.aspx?thing=346464646
[download]

The $^a and $^b in the map block tell it to take two elements at a time, while the X in the parenthesis tells it to cross join the two arrays. Very very handy for this kind of thing.

Such cross-like operations are common enough that there is also a cross meta-operator in Perl 6 to trick an ordinary binary operator into doing it. So instead of saying:

    my @combined = (@urls X @ids).map: {$^a ~ $^b};
[download]

you can just say

    my @combined = @urls X~ @ids;
[download]

That is, just put X in front of the ordinary operator to make it a cross operator.

Update: changed old X~X form to new X~ form.

Interesting. But, unfortunately I am using perl 5.8. Where can I read about the new arrivals in perl 6 version?

http://www.perlfoundation.org/perl6/index.cgi is a pretty good place to start reading about it.


___________
Eric Hodges