sub a:b['c'] generates .sub "a:b<Capture[0x...]>"

[Benabik]

It seems that colonpair_str returns what I guess to be a PCT node instead of a string for the :b['c'] syntax. To expand the terse example in the title:

sub a:b['c']() { say("hi") }

generates the following PIR

.sub "a:b<Capture[0xeafa6588]>"  :subid("11_1313977936.07837") :outer("10_1313977936.07837")
    $P100 = "say"("hi")
    .return ($P100)
.end

This may seem somewhat cosmetic, but this also affects grammars, so the following from nom's setting:

    token special_variable:sym['$<'] {
        <sym> <!before \s* \w+ \s* '>' >

generates:

.sub "special_variable:sym<Capture[0xf4820400]>"  :subid("461_1313948700.73446") :method :outer("11_1313948700.73446")
    # cut for brevity
    ne $S10, "Capture[0xf4820400]", rx2590_fail

Which means that it doesn't match what it should.

[arnsholt]

Latest NQP generates .sub "a:b<c>".