From mark@kli.org

 
 
 
  $ perl6 --version
This is perl6 version 2011.10-98-g1985138 built on parrot 3.9.0 revision RELEASE_3_9_0-51-g65e6ab7
$ perl6

"e u a u e u a" ~~ /​:sigspace ("a") ([<[aeiou]> ]+)$/
=> < a u e u a>
0 => <a>
1 => <u e u a>

"e u a u e u a" ~~ /​:sigspace .("a") ([<[aeiou]> ]+)$/
#<failed match>
# Only a single . was added before the ("a"), which is plainly not at the beginning of the line.
"e u a u e u a" ~~ /​:sigspace .?("a") ([<[aeiou]> ]+)$/
=> < a u e u a>
0 => <a>
1 => <u e u a>

# OK, making the . optional works, but what if I try to capture it?
"e u a u e u a" ~~ /​:sigspace (.?)("a") ([<[aeiou]> ]+)$/
#<failed match>
# Surely choosing to capture a part of the match shouldn't affect whether it matches at all!

This all works differently if we use a space between the . and the ("a"). It looks like
at least some my own difficulty in the particular case relates to the space after
the :sigspace adverb counting as whitespace, but does that explain everything?

~mark

From @moritz

A shorter way to reproduce the same thing​:

"u a" ~~ /​:sigspace .'a'/
#<failed match>

Or even

"u " ~~ /<.ws>.$/
#<failed match>

So, seems to be a problem with <.ws>, and not captures or backtracking
(at least not anymore :-)

@moritz - Status changed from 'new' to 'open'

From @moritz

On second thought, it's not even a bug​: <.ws> doesn't backtrack. It
matches the space on the first pass through, and doesn't give it up
anymore, so the . has nothing to match.

@moritz - Status changed from 'open' to 'rejected'