[jbert]

I'm interested in the details of the variable scoping problems are. I'd have thought that perl's explicit declaration of lexicals with 'my' was better than "declare at first use" of javascript/python/ruby.

[There's a class of mistake where you typo a variable name which you can make in those languages which gets caught in a language where you have to declare variables.]

[SwellJoe]

I share your confusion over that assertion. I actually miss Perl's excellent block scope capabilities when working in almost every other similar language. I imagine most other languages have gotten it, by now, but a few years ago when I was using Python for a few years for a client it didn't have block scope and I missed it. A brief googling actually hints that maybe Python still doesn't have block scope, which is pretty weird, so I assume I'm googling wrong.

But, perhaps discussion of scope problems in Perl is just when talking about very old code. Perl had some scope atrocities back in Perl 4 and some weirdness in early Perl 5 days. But, for the past 15 years or so, it's been very predictable and has improved (like the syntactic sugar of "my" declarations in foreach or while expressions).

[leoc]

I seem to recall that Python not only didn't (doesn't?) have block scope, it tended (tends?) to produce a completely, blithely cryptic error message when the programmer assumes that it does have proper block scope and writes a program that fails as a result. Discovering that was memorable.

[fanf2]

One weirdness I encountered recently with nested functions:

  sub a {
    my $v;
    sub b {
      # captures only the first
      # instance of $v
    }
  }

This is because named subroutines are created once and variable captures are resolved at that time. I expected the more normal capture semantics you get with anonymous subs, like

  sub a {
    my $v;
    my $b = sub {
      # a new $b each time
      # captures each $v
    }
  }