"Iteration Past End" with Empty Character Class that has unescaped space in it

[p6rt]

Migrated from rt.perl.org#130586 (status was 'resolved')

Searchable as RT130586$

[p6rt]

From @zoffixznet

the :sigspace adverb does not affect spaces inside character class. Should it?

Note that having JUST space inside also causes a weird error to be thrown​:

<Zoffix> m​: say " " ~~ m​:s/<-[ ]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«===SORRY!===␤Iteration past end of iterator␤»

<Zoffix> m​: say " " ~~ m​:s/<-[x ]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
<Zoffix> m​: say " " ~~ m​:s/<-[ x]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
<Zoffix> m​: say " " ~~ m​:sigspace/<-[ x]>/

<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
<Zoffix> m​: say " " ~~ m​:sigspace/<[x ]>/

Discovery​: https://irclog.perlgeek.de/perl6/2017-01-19#i_13948706

[p6rt]

From @jnthn

On Wed, 18 Jan 2017 18​:16​:20 -0800, cpan@​zoffix.com wrote​:

the :sigspace adverb does not affect spaces inside character class.
Should it?

No, I don't think so. `<.ws>` can match any number of characters (including zero). A char class is defining something a matcher for a single character.

Note that having JUST space inside also causes a weird error to be
thrown​:

<Zoffix> m​: say " " ~~ m​:s/<-[ ]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«===SORRY!===␤Iteration past end
of iterator␤»

That one's certainly a bug. Since whitespace in a charclass is not significant, then it should complain about an empty charclass, preferably with a hint about space.

<Zoffix> m​: say " " ~~ m​:s/<-[x ]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
This matches because it's the same as m​:s/<-[x]>/. You'd need to write <-[x\ ]> to make the space significant.

<Zoffix> m​: say " " ~~ m​:s/<-[ x]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
Ditto.

/jnthn

[p6rt]

The RT System itself - Status changed from 'new' to 'open'

[p6rt]

From @AlexDaniel

Further discussion on rakudo/rakudo#1622.

On 2017-01-21 14​:13​:42, jnthn@​jnthn.net wrote​:

On Wed, 18 Jan 2017 18​:16​:20 -0800, cpan@​zoffix.com wrote​:

the :sigspace adverb does not affect spaces inside character class.
Should it?

No, I don't think so. `<.ws>` can match any number of characters
(including zero). A char class is defining something a matcher for a
single character.

Note that having JUST space inside also causes a weird error to be
thrown​:

<Zoffix> m​: say " " ~~ m​:s/<-[ ]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«===SORRY!===␤Iteration past end
of iterator␤»

That one's certainly a bug. Since whitespace in a charclass is not
significant, then it should complain about an empty charclass,
preferably with a hint about space.

<Zoffix> m​: say " " ~~ m​:s/<-[x ]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
This matches because it's the same as m​:s/<-[x]>/. You'd need to write
<-[x\ ]> to make the space significant.

<Zoffix> m​: say " " ~~ m​:s/<-[ x]>/
<camelia> rakudo-moar 1aeea1​: OUTPUT«｢ ｣␤»
Ditto.

/jnthn

[p6rt]

@AlexDaniel - Status changed from 'open' to 'resolved'