##
## Sigil invariance
##

array = [1, 3, 5, 12, 37, 42]

# I use symbols here, but one can use strings: {'alpha' => 4, 'beta' =
+> 6}
# 
hash = {:alpha => 4, :beta => 6}

third = array[2]

# This would be hash['beta'] when using strings
#
beta = hash[:beta]

# Ruby doesn't slice as a part of syntax - there are special member 
# methods for that. I personally see it as a good thing.
odds = array.values_at(1, 3, 5)
bets = hash.values_at(:alpha, :beta)
[download]

You like having symbol names automagically spring into existence without even the namespacing protection of different prefixes?

...spring into existence without even the namespacing protection of different prefixes?

The extremely miniscule benefit of having same-named variables of different types is one I would gladly trade for the reduction in typing and visual clutter. Some people code according to style guidelines which explicitly discourage the use of this feature — and for good reason, imho.

Note that in a name like $x, the keystrokes involved are three times that necessary to type x. That really belies the alleged priority given to Huffman coding in the design of Perl.

Of course one will argue that good Perl coders don't use names that short; but a keystroke penalty against them is not a good way to encourage better behavior. The two keystoke penalty is incurred on every variable name occurrence.

I think it also worth mentioning that the elimination of sigils would have major PR benefit as well.

As for springing into existence — this (undeclared variables) is the sort of behavior that the interpreter should be in the business of discouraging. The penalty of having to declare a variable is incurred only once per variable.

A word spoken in Mind will reach its own level, in the objective world, by its own weight