From @masak

<moritz> m​: -> $f, $s { return $f + $s }(23, 42)
<camelia> rakudo-moar 11afe1​: ( no output )
<moritz> m​: say -> $f, $s { return $f + $s }(23, 42)
<camelia> rakudo-moar 11afe1​: ( no output )
<moritz> m​: say (-> $f, $s { return $f + $s })(23, 42)
<camelia> rakudo-moar 11afe1​: ( no output )
<moritz> ffs?
<moritz> m​: say 23 + 42
<camelia> rakudo-moar 11afe1​: OUTPUT«65␤»
<FROGGS> m​: say( -> $f, $s { return $f + $s }(23, 42) )
<camelia> rakudo-moar 11afe1​: ( no output )
<FROGGS> err
<moritz> oh
<FROGGS> a return in a pointy >.<
<moritz> return inside a lambda is a BAD IDEA
<moritz> return is only for routines
<moritz> it should say "return without a routine"
<FROGGS> if it does not do that it should state that the
return-exception missed its handler
<masak> m​: return; say "alive"
<camelia> rakudo-moar 11afe1​: ( no output )
<masak> I'd go so far as to consider that a bug.
<FROGGS> yeah
<masak> it can be caught *statically*.
<PerlJam> agreed
* masak submits rakuobug
<moritz> since control exceptions are lexotic, and all that
<masak> `return` in the mainline is always wrong.
<PerlJam> masak​: sub MAIN { return 42; } # ;-)
<masak> PerlJam​: that's un-wrong but meaningless :)
<PerlJam> masak​: it's a gentle nudge toward improving the precision of
our words :)
<masak> PerlJam​: I stand by what I said. MAIN is a sub, so `return` is
fine there.
<masak> PerlJam​: it's only not fine outside of any routine.
<FROGGS> "Can only return from within a method or sub"
<masak> FROGGS​: "routine"
<masak> that's the Perl 6 word for it.
<masak> FROGGS​: you can return from a macro, too. for example.

From @pmichaud

Per further discussion on #perl6...

By definition, I think that "lexotic" sharply limits what can be done statically. If a C<return> (inside of a block) isn't lexically scoped, it's dynamically scoped. About the only place it could be caught statically is as a bare 'return' inside of a mainline (and even there, I'm wondering if there is a case where a mainline might dynamically be invoked from inside of a Routine).

However, a return that is invoked dynamically outside of any Routine should throw a runtime error, so I'm changing the topic of this ticket to match that.

Pm

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

From @lizmat

[14​:22​:01] <lizmat> m​: my $a = do { return 42 }
[14​:22​:02] <+camelia> rakudo-moar 5f9cd5​: ( no output )
[14​:22​:34] <lizmat> $ MVM_SPESH_DISABLE=1 perl6 -e 'my $a = do { return 42 }'
[14​:22​:34] <lizmat> Segmentation fault​: 11
[14​:22​:41] <lizmat> segfaults locally for me
[14​:22​:54] <lizmat> can someone confirm ?
[14​:23​:29] <colomon> seems to throw me in the REPL?!?
[14​:23​:41] <colomon> lizmat​: ^^
[14​:23​:57] <colomon> ah, no, I see
[14​:24​:04] <colomon> seg fault on OS X
[14​:24​:14] <colomon> and linux
[14​:24​:27] <lizmat> also in the REPL
[14​:25​:25] <colomon> lizmat​: wasn't actually the REPL, I missed the closing quote in my copy-n-paste so the shell wanted more input.
[14​:25​:51] <lizmat> actually, I can golf it down to just { return }
[14​:26​:01] <lizmat> $ MVM_SPESH_DISABLE=1 perl6 -e '{ return }'
[14​:26​:01] <lizmat> Segmentation fault​: 11
[14​:26​:24] <lizmat> hmmm... actually just return is enough
[14​:27​:04] lizmat rakudobugs
[14​:27​:18] <colomon> lizmat​: confirmed for just { return }, too
[14​:27​:32] <lizmat> $ perl6-p -e 'return'
[14​:27​:32] <lizmat> Attempt to return outside of any Routine
[14​:27​:39] <lizmat> is what it probably should be

From @smls

A `return` outside a routine now always seems to throw a run-time error, as it should​:

  $ perl6 -e 'return'
  Attempt to return outside of any Routine
  in block <unit> at -e line 1

  $ perl6 -e '{ return }'
  Attempt to return outside of any Routine
  in block <unit> at -e line 1

  $ MVM_SPESH_DISABLE=1 perl6 -e '{ return }'
  Attempt to return outside of any Routine
  in block <unit> at -e line 1

Bisectable points to this fix from Dec 2015​:

  https://gist.github.com/Whateverable/63b5da597e229d167d8708a43d4501de
  rakudo/rakudo@a4ca12a

This ticket can be closed once tests are added to roast.

From @skids

On Sun, 10 Sep 2017 11​:22​:29 -0700, smls75@​gmail.com wrote​:

A `return` outside a routine now always seems to throw a run-time
error, as it should​:

$ perl6 -e 'return'
Attempt to return outside of any Routine
in block <unit> at -e line 1

$ perl6 -e '{ return }'
Attempt to return outside of any Routine
in block <unit> at -e line 1

$ MVM_SPESH_DISABLE=1 perl6 -e '{ return }'
Attempt to return outside of any Routine
in block <unit> at -e line 1

Bisectable points to this fix from Dec 2015​:

https://gist.github.com/Whateverable/63b5da597e229d167d8708a43d4501de
rakudo/rakudo@a4ca12a

This ticket can be closed once tests are added to roast.

It looks to me like the tests already in S04-statements/return.t should suffice, no?

(This appears to be another case where a bug filed and fixed during the
craziness of the 2015 release and GLR was not followed up in RT
by closing the ticket.)

Since MVM_* environment variables are not part of the Perl 6 specification,
tests for those cases aren't a matter for roast... possibly they could
go in the rakudo tree, but it would probably be better to figure out how to golf
that SEGV down into an nqp one-liner... rakudo can no longer produce the
situation that caused that SEGV, so... where to follow that thread I dunno.

From @smls

It looks to me like the tests already in S04-statements/return.t
should suffice, no?

You're right. Closing the ticket.

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