From @masak

<masak> has Grammar​::Debugger bitrotted?
https://gist.github.com/masak/d9bd70d9f196a39d30bb
* masak attempts to diagnose that error
<Ven> has $!auto-continue = False;
<masak> shouldn't be immutable.
<masak> that error looks very bogus.
<masak> let's see what I find during the golf after the update.
<masak> just upgraded my Rakudo. Grammar​::Debugger error persists.
* masak enjoys golfing something again
<masak> oh, "interesting" :/
<masak> it's related to publish_method_cache
<FROGGS> uhh
<masak> I think I've uncovered a second error under the first error.
<masak> maybe this will be a bigger clue for someone else than it is for me.
<masak> https://gist.github.com/masak/d9bd70d9f196a39d30bb updated
with the underlying error.
<masak> I think the original error is caused as part of reporting the
second one.
<lizmat> ah, maybe it is related to jnthn optimization where $_, $/
and friends are only allocated when used
<masak> https://gist.github.com/masak/d9bd70d9f196a39d30bb updated
with the golfiest version of Grammar​::Debugger that still triggers the
error.
<lizmat> and *that* may be related to the async $_ re-use issue
<masak> jnthn​: ^ when you have time, you definitely should have a look at this.
<masak> jnthn​: near as I can tell, at some point Grammar​::Debugger is
trying to call .TOP on the wrong object.
<masak> lizmat​: optimization and $/ sounds eerily correct.
<masak> lizmat​: since 8584dbea is now the prime suspect, I'm going to
build with 8584dbea^ and see if the error goes away. then with
8584dbea and see if it comes back.
<masak> another data point​: problem persists in a Rakudo from 2014-01-31.
<masak> so jnthn++'s recent optimizations are cleared of suspicion.
<masak> going back another year to try to build there.
<masak> (how long has Grammar​::Debugger been busted without anyone noticing?)
<masak> star​: my $*IN = role { method get() { "" } }; use
Grammar​::Debugger; grammar G { regex TOP { hi } }; G.parse("hi")
<camelia> star-p 2014.04​: OUTPUT«TOP␤> Cannot modify an immutable value [...]
<camelia> ..star-m 2014.04​: OUTPUT«TOP␤> No such method 'STORE' for
invocant of type 'Bool' [...]
<masak> ha!
* masak submits rakudobug
<masak> interesting, I thought this was a moar-only problem. but
that's the same issue manifesting slightly differently on two VMs.
<masak> ok, doing a bisect using parrot as a backend.

From @perlpilot

On Mon May 26 09​:53​:30 2014, masak wrote​:

<masak> has Grammar​::Debugger bitrotted?
https://gist.github.com/masak/d9bd70d9f196a39d30bb
* masak attempts to diagnose that error
<Ven> has $!auto-continue = False;
<masak> shouldn't be immutable.
<masak> that error looks very bogus.
<masak> let's see what I find during the golf after the update.
<masak> just upgraded my Rakudo. Grammar​::Debugger error persists.
* masak enjoys golfing something again
<masak> oh, "interesting" :/
<masak> it's related to publish_method_cache
<FROGGS> uhh
<masak> I think I've uncovered a second error under the first error.
<masak> maybe this will be a bigger clue for someone else than it is
for me.
<masak> https://gist.github.com/masak/d9bd70d9f196a39d30bb updated
with the underlying error.
<masak> I think the original error is caused as part of reporting the
second one.
<lizmat> ah, maybe it is related to jnthn optimization where $_, $/
and friends are only allocated when used
<masak> https://gist.github.com/masak/d9bd70d9f196a39d30bb updated
with the golfiest version of Grammar​::Debugger that still triggers the
error.
<lizmat> and *that* may be related to the async $_ re-use issue
<masak> jnthn​: ^ when you have time, you definitely should have a look
at this.
<masak> jnthn​: near as I can tell, at some point Grammar​::Debugger is
trying to call .TOP on the wrong object.
<masak> lizmat​: optimization and $/ sounds eerily correct.
<masak> lizmat​: since 8584dbea is now the prime suspect, I'm going to
build with 8584dbea^ and see if the error goes away. then with
8584dbea and see if it comes back.
<masak> another data point​: problem persists in a Rakudo from 2014-01-
31.
<masak> so jnthn++'s recent optimizations are cleared of suspicion.
<masak> going back another year to try to build there.
<masak> (how long has Grammar​::Debugger been busted without anyone
noticing?)
<masak> star​: my $*IN = role { method get() { "" } }; use
Grammar​::Debugger; grammar G { regex TOP { hi } }; G.parse("hi")
<camelia> star-p 2014.04​: OUTPUT«TOP␤> Cannot modify an immutable
value [...]
<camelia> ..star-m 2014.04​: OUTPUT«TOP␤> No such method 'STORE' for
invocant of type 'Bool' [...]
<masak> ha!
* masak submits rakudobug
<masak> interesting, I thought this was a moar-only problem. but
that's the same issue manifesting slightly differently on two VMs.
<masak> ok, doing a bisect using parrot as a backend.

Has this been fixed? I can't reproduce it on​:

This is perl6 version 2014.08-360-ge99b04b built on MoarVM version 2014.08-139-gb37f93e

-Scott (PerlJam)

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

From @usev6

On Tue Sep 16 06​:33​:06 2014, duff wrote​:

[...]
Has this been fixed? I can't reproduce it on​:

This is perl6 version 2014.08-360-ge99b04b built on MoarVM version
2014.08-139-gb37f93e

-Scott (PerlJam)

I also can't reproduce the error with a current Rakudo (on Moar, Parrot and JVM)​:

$ PERL6LIB=lib perl6 -e 'my $*IN = role { method get() { "" } }; use Grammar​::Debugger; grammar G { regex TOP { hi } }; G.parse("hi")'
TOP

* MATCH "hi"

I'm closing this ticket now.

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