From @masak

<masak> rakudo​: $_ = 'foo'; s​:g/ /a/; say 'alive'
<p6eval> rakudo 8681da​: OUTPUT«alive␤»
<masak> std​: $_ = 'foo'; s​:g/ /a/
<p6eval> std 625303c​: OUTPUT«�[31m===�[0mSORRY!�[31m===�[0m␤Null
pattern not allowed [...] FAILED 00​:01 122m␤»
* masak submits rakudobug

From anteusz@freemail.hu

Hi,

Here is the patch..

Jaffa4

From anteusz@freemail.hu

Grammar.pm

From @moritz

And as an actual patch​:

--- a/src/Perl6/Grammar.pm
+++ b/src/Perl6/Grammar.pm
@​@​ -1477,6 +1477,7 @​@​ token quote​:sym<s> {
  }
  <.setup_quotepairs>
  [
+ | '/' \s* '/' <.panic​: "Null regex in substitution not allowed">
  | '/' <p6regex=.LANG('Regex','nibbler')> <?[/]> <quote_EXPR​:
'​:qq'> <.old_rx_mods>?
  | '[' <p6regex=.LANG('Regex','nibbler')> ']'
  <.ws> [ '=' || <.panic​: "Missing assignment operator"> ]

But I fear this is the wrong approach. Instead of doing a separate null
pattern check in every quoting construct, the check should be done in
the regex parsing code once, as STD.pm6 does it.

But I'll leave the final say to pmichaud or jnthn.

Cheers,
Moritz

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

From @pmichaud

On Mon, Feb 07, 2011 at 11​:36​:22AM +0100, Moritz Lenz wrote​:

+ | '/' \s* '/' <.panic​: "Null regex in substitution not allowed">

But I fear this is the wrong approach. Instead of doing a separate
null pattern check in every quoting construct, the check should be
done in the regex parsing code once, as STD.pm6 does it.

I completely agree; the check for null regex needs to be done in the
regex parsing code. In particular, the following regex should also
produce a "null pattern" exception of some sort, which the patch
doesn't address​:

  / abc | | def /

Patch rejected; thanks for submitting!

Pm

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

From anteusz@freemail.hu

2011.02.07. 11​:36 keltezéssel, Moritz Lenz via RT írta​:

And as an actual patch​:

--- a/src/Perl6/Grammar.pm
+++ b/src/Perl6/Grammar.pm
@​@​ -1477,6 +1477,7 @​@​ token quote​:sym<s> {
}
<.setup_quotepairs>
[
+ | '/' \s* '/'<.panic​: "Null regex in substitution not allowed">
| '/'<p6regex=.LANG('Regex','nibbler')> <?[/]> <quote_EXPR​:
'​:qq'> <.old_rx_mods>?
| '['<p6regex=.LANG('Regex','nibbler')> ']'
<.ws> [ '=' ||<.panic​: "Missing assignment operator"> ]

But I fear this is the wrong approach. Instead of doing a separate null
pattern check in every quoting construct, the check should be done in
the regex parsing code once, as STD.pm6 does it.

But I'll leave the final say to pmichaud or jnthn.

Cheers,
Moritz

Note , this approach has been used already in the same grammar.pm. If
you search for Null regex, you will find it.

Jaffa4

From @coke

On Wed Jan 12 09​:29​:31 2011, masak wrote​:

<masak> rakudo​: $_ = 'foo'; s​:g/ /a/; say 'alive'
<p6eval> rakudo 8681da​: OUTPUT«alive␤»
<masak> std​: $_ = 'foo'; s​:g/ /a/
<p6eval> std 625303c​: OUTPUT«�[31m===�[0mSORRY!�[31m===�[0m␤Null
pattern not allowed [...] FAILED 00​:01 122m␤»
* masak submits rakudobug

This now dies correctly​:

19​:46 < [Coke]> m​: $_ = 'foo'; s​:g/ /a/; say 'alive'
19​:46 <+camelia> rakudo-moar 974d00​: OUTPUT«===SORRY!=== Error while compiling
  /tmp/LZQwMHpP5h␤Null regex not allowed␤at
  /tmp/LZQwMHpP5h​:1␤------> $_ = 'foo'; s​:g/ ⏏/a/; say 'alive'␤
  expecting any of​:␤ colon pair (restricted)␤»

Closable with tests.
--
Will "Coke" Coleda

From @moritz

On Tue Feb 25 16​:46​:33 2014, coke wrote​:

On Wed Jan 12 09​:29​:31 2011, masak wrote​:

<masak> rakudo​: $_ = 'foo'; s​:g/ /a/; say 'alive'
<p6eval> rakudo 8681da​: OUTPUT«alive␤»
<masak> std​: $_ = 'foo'; s​:g/ /a/
<p6eval> std 625303c​: OUTPUT«�[31m===�[0mSORRY!�[31m===�[0m␤Null
pattern not allowed [...] FAILED 00​:01 122m␤»
* masak submits rakudobug

This now dies correctly​:

19​:46 < [Coke]> m​: $_ = 'foo'; s​:g/ /a/; say 'alive'
19​:46 <+camelia> rakudo-moar 974d00​: OUTPUT«===SORRY!=== Error while
compiling
/tmp/LZQwMHpP5h␤Null regex not allowed␤at
/tmp/LZQwMHpP5h​:1␤------> $_ = 'foo'; s​:g/ ⏏/a/; say
'alive'␤
expecting any of​:␤ colon pair (restricted)␤»

Closable with tests.

We have tests now​:

commit fb9afbd6a49546a69b002816da99e0d0b5a0afc8
Author​: Moritz Lenz <mlenz@​noris.net>
Date​: Wed Feb 26 09​:24​:27 2014 +0100

  Test for RT #​82142

diff --git a/S32-exceptions/misc.t b/S32-exceptions/misc.t
index 1dd6284..52938e6 100644
--- a/S32-exceptions/misc.t
+++ b/S32-exceptions/misc.t
@@ -142,6 +142,8 @@ throws_like q[/ a & /], X::Syntax::Regex::NullRegex;
 # RT 67554
 throws_like q{/ [] /}, X::Syntax::Regex::NullRegex;
 throws_like q{/ | /}, X::Syntax::Regex::NullRegex;
+# RT #82142
+throws_like q{s//b/}, X::Syntax::Regex::NullRegex;
 
 
 throws_like 'sub f($a?, $b) { }', X::Parameter::WrongOrder,

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