Parse error on a way of writing ranges in infix:<**> quantifiers that probably should work in Rakudo

[p6rt]

Migrated from rt.perl.org#122502 (status was 'resolved')

Searchable as RT122502$

[p6rt]

From @masak

<psch> m​: /a ** 1 ..2/ # this does confuse me.
<camelia> rakudo-moar 5d5ec1​: OUTPUT«===SORRY!===␤Cannot find method 'ast'␤»
* masak submits rakudobug
<psch> the curious bit is that P6Regex​::Grammar has <.integer> '..'
<.throw_bare_spaces_in_range>
<psch> which is the patch jnthn merged a bit
<psch> but apparently it matches differently and gets through to the
corresponding action in Perl 6's Actions.nqp
<psch> where there's nothing in $<min>
<psch> because it shouldn't have matched or somesuch...?

It's not clear to me whether the above should just work (probably) or
give a better error message. But it should do either of those.

[p6rt]

From @peschwa

On Sun Aug 10 09​:00​:08 2014, masak wrote​:

<psch> m​: /a ** 1 ..2/ # this does confuse me.
<camelia> rakudo-moar 5d5ec1​: OUTPUT«===SORRY!===␤Cannot find method 'ast'␤»
* masak submits rakudobug
<psch> the curious bit is that P6Regex​::Grammar has <.integer> '..'
<.throw_bare_spaces_in_range>
<psch> which is the patch jnthn merged a bit
<psch> but apparently it matches differently and gets through to the
corresponding action in Perl 6's Actions.nqp
<psch> where there's nothing in $<min>
<psch> because it shouldn't have matched or somesuch...?

It's not clear to me whether the above should just work (probably) or
give a better error message. But it should do either of those.

It should not work.

21​:33 <psch> std​: /m ** 1 ..1/
21​:33 <camelia> std 0f2049c​: OUTPUT<<===SORRY!===NLSpaces not allowed in bare range at /tmp/djAk9z1rR0 line 1​:NL------> /m ** 1
  ..?1/NLParse failedNLFAILED 00​:01 122mNL>>

X​::Syntax​::Regex​::SpacesInBareRange has recently been created, but the patch which did so wasn't submitted for merging before changes to how the range is parsed broke an assumption between P6Regex​::Grammar and P6Regex​::Actions. A PR for NQP exists at [0].

[0] Raku/nqp#174

[p6rt]

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

[p6rt]

From @peschwa

Am So 10. Aug 2014, 12​:40​:20, peschwa@​gmail.com schrieb​:

On Sun Aug 10 09​:00​:08 2014, masak wrote​:

<psch> m​: /a ** 1 ..2/ # this does confuse me.
<camelia> rakudo-moar 5d5ec1​: OUTPUT«===SORRY!===␤Cannot find method
'ast'␤»
* masak submits rakudobug
<psch> the curious bit is that P6Regex​::Grammar has <.integer> '..'
<.throw_bare_spaces_in_range>
<psch> which is the patch jnthn merged a bit
<psch> but apparently it matches differently and gets through to the
corresponding action in Perl 6's Actions.nqp
<psch> where there's nothing in $<min>
<psch> because it shouldn't have matched or somesuch...?

It's not clear to me whether the above should just work (probably) or
give a better error message. But it should do either of those.

It should not work.

21​:33 <psch> std​: /m ** 1 ..1/
21​:33 <camelia> std 0f2049c​: OUTPUT<<===SORRY!===NLSpaces not allowed
in bare range at /tmp/djAk9z1rR0 line 1​:NL------> /m ** 1
..?1/NLParse failedNLFAILED 00​:01 122mNL>>

X​::Syntax​::Regex​::SpacesInBareRange has recently been created, but the
patch which did so wasn't submitted for merging before changes to how
the range is parsed broke an assumption between P6Regex​::Grammar and
P6Regex​::Actions. A PR for NQP exists at [0].

[0] Raku/nqp#174

The mentioned NQP-PR has been merged, a test has been added and corrected in roast commits 59436af1 and a0a80b76.

[p6rt]

From @usev6

Since the test in S32-exceptions/misc.t passes for Rakudo on Moar, Parrot and JVM, I'm closing this ticket.

[p6rt]

From @usev6

Since the test in S32-exceptions/misc.t passes for Rakudo on Moar, Parrot and JVM, I'm closing this ticket.

[p6rt]

@usev6 - Status changed from 'open' to 'resolved'