I'm not saying that the design of the P6 builtin is ST, just that it is functionally equivalent, and even more concise.

And it's not just about the "decorate" step, which your "key extractor" closure is roughly equivalent of.

One critical point: while a "key extractor" closure is indeed roughly equivalent to the decorate step in terms of the role it's playing, it's multiple key extractor closures that is basically fully equivalent to the before map.

Because the ST has an explicit sort block (normally, unless it's the GRT variant :-)), it allows to execute arbitrarily complex Perl code in the comparison. Where does one do that in this Perl 6 "equivalent"?

You just specify a comparator closure (or multiple, if you want).

(P6 can tell the difference due to a closure's number of args. If it has one, it's a key extractor closure, if it has two it's a comparator closure.)

For lots more examples which are still correct modulo a few minor syntax shifts, see "The Sort Problem: a definitive ruling" from 2004.

Which got me to focus on the Pair variant which maybe helps clarify the connection for you:

sort { #`( key extractor closure) } => { #`( comparator closure) },
     { #`( key extractor closure) } => { #`( comparator closure) },
     ...
     @array
[download]