What would you expect this to emit?
my $qr = qr/f/; print qr/[#]$qr/x;
my $qr = qr/(f)/; print "#f" =~ m/[#](f)/x; print "#f" =~ m/[#]$qr/x;
looks like a bug to me. somehow the interpolation stops at the first # found.
% steph@ape (/home/stephan) %
% cat t1.px
#! /usr/bin/perl
my $qr = qr/f/;
print qr/$qr[#]$qr/x;
my $qr2 = qr/f/;
print qr/[b]$qr2/x;
% steph@ape (/home/stephan) %
% perl -w t1.px
(?x-ism:(?-xism:f)[#]$qr)(?x-ism:[b](?-xism:f))
Re:seems to be a bug
Aristotle on 2007-09-21T13:55:47
No, it’s not really a bug. An unescaped hashmark always starts a comment in patterns where
/xapplies. If the hashmark were backslash, it would produce the result that Chris probably expected.There is a definite argument to make that it’s counterintuitive, though. Whitespace in character classes is significant even under
/x, so one would expect comments (which are usually thought of as a special kind of ignorable whitespace) would not be allowed in the middle of a character class.Re:seems to be a bug
ChrisDolan on 2007-09-22T01:12:45
If it's not a bug, then why does the following print "f" and not "1"?Or more to the point, if the "#" is a comment character, why is it not a syntax error for an unclosed "[]" group?print "#f" =~ m/[#](f)/x;Re:seems to be a bug
Aristotle on 2007-09-22T07:13:03
Ack. Yeah, definitely a bug then. Seems different parts of the regex parser have different ideas of when they’re inside a comment, or something like that.
Did you write to p5p?
Re:seems to be a bug
ChrisDolan on 2007-09-22T11:31:07
My first attempt, before posting this journal entry, vanished into the ether (in retrospect, probably because my ISP blocks port 22). I just re-submitted. Thanks for reminding me.