Reply to thread

Message

[QUOTE="Uri Guttman, post: 4863076"]
JG> my @ks = qw(two three one four);
JG> my %h = ( one => 1, three => 333 );

JG> my @sorted = (
JG>   (sort {length $h{$a} <=> length $h{$b}} grep {exists $h{$_}} @ks),
JG>   (grep {!exists $h{$_}} @ks)
JG> );

all this confusing redundant code is annoying me. i left an exercise on
doing this all one time but it seems i will have to answer it myself. i
will assume length of a key will always be less than some large number
which is how missing keys will sort late. this is an untested ST but
Sort::Maker can generate this for you too.

@sorted = map $_->[0],
sort { $a->[1] <=> $b->[1] }
map [ $_, exists($h{$_}) ? length $h{$_} : 9999999 ],
@unsorted ;

there is only one call to exists and length and it is much clearer what
you mean to do. and Sort::Maker cleans up all the cruft and you only
need the exists expression.

the point is that you want to do key extraction/processing only one time
both for speed and for clarity. perl6's sort supports this style and
drops the $a/$b redundant stuff in general.

uri
[/QUOTE]

Verification