On Thu, Aug 22, 2019 at 6:11 PM yary <not....@gmail.com> wrote: > Perl 6 is doing the right thing. The dot matches any character. In > this case, matching the final ':'. The next bit of the regex says the > cursor has to be after 1:, and indeed, after matching the ':' the > cursor is after '1:', so the substitution succeeds. >
My real use case, that I tried to provide a simplified example of, was to process some pretty-printed JSON. Less simplified this time, I wanted to change all "foo": "whatever" strings to "foo": "*". In Perl 5 I would have done: s/(?<="foo": ")[^"]+/*/; Trying to express this in Perl 6, I thought "lookbehind" would naturally translate to a "before" assertion: s/<?before '"foo": "'><-["]>+/*/; ...but that didn't work. Various other attempts led to the simplified example I originally provided. Long story short, it seems that a Perl 5 (?<=...) lookbehind translates to a Perl 6 <?after ...> assertion, and likewise a Perl 5 (?=...) lookahead translates to a Perl 6 <?before ...> assertion. The terminology just confused me due to my prior Perl 5 experience.