From @lizmat

[23​:21​:43] <lizmat> m​: sub aa { for @​_ { .say } }; aa 42
[23​:21​:43] <+camelia> rakudo-moar 009400​: OUTPUT«42␤»
[23​:21​:55] <lizmat> m​: sub aa { if @​_ { for @​_ { .say } } }; aa 42
[23​:21​:55] <+camelia> rakudo-moar 009400​: ( no output )
[23​:22​:15] <lizmat> m​: sub aa { if @​_ { say @​_ } }; aa 42
[23​:22​:15] <+camelia> rakudo-moar 009400​: OUTPUT«[]␤»
[23​:22​:33] <lizmat> seems all wrong to me ?
[23​:24​:39] <lizmat> looks like an "if @​_" eats @​_ ?
[23​:27​:42] <grondilu> m​: { @​_ && { .say for @​_ } }(42)
[23​:27​:42] <+camelia> rakudo-moar 009400​: ( no output )
[23​:28​:09] <grondilu> m​: { @​_ && "ok" }(42)
[23​:28​:09] <+camelia> rakudo-moar 009400​: ( no output )
[23​:28​:17] <grondilu> m​: say { @​_ && "ok" }(42)
[23​:28​:18] <+camelia> rakudo-moar 009400​: OUTPUT«ok␤»
[23​:28​:35] <PerlJam> m​: sub foo { if @​_ { say "hi​: @​_[]" } }; foo 42;
[23​:28​:36] <+camelia> rakudo-moar 009400​: OUTPUT«hi​: ␤»
[23​:28​:59] <grondilu> m​: say { @​_ && @​_ }(42)
[23​:28​:59] <+camelia> rakudo-moar 009400​: OUTPUT«[42]␤»
[23​:29​:14] <PerlJam> m​: sub foo { say @​_.WHAT }; foo 42;
[23​:29​:14] <+camelia> rakudo-moar 009400​: OUTPUT«(Array)␤»
[23​:30​:01] <grondilu> m​: { @​_ && { .say for @​_ }() }(42)
[23​:30​:02] <+camelia> rakudo-moar 009400​: ( no output )
[23​:30​:09] <grondilu> m​: { @​_ && { .say for @​_ }(@​_) }(42)
[23​:30​:10] <+camelia> rakudo-moar 009400​: OUTPUT«42␤»
[23​:30​:37] grondilu stops pretending he has an idea of what's going on there
[23​:32​:05] <PerlJam> m​: sub foo { say @​_; if @​_ { say @​_ } }; foo 42;
[23​:32​:05] <+camelia> rakudo-moar 009400​: OUTPUT«[42]␤[]␤»
[23​:32​:23] <PerlJam> I think it's some kind of iterator problem.
[23​:38​:00] <PerlJam> m​: sub foo { say @​_; if @​_ { say @​_; }; say @​_ }; foo 42;
[23​:38​:00] <+camelia> rakudo-moar 009400​: OUTPUT«[42]␤[]␤[42]␤»
[23​:38​:30] <PerlJam> or ... maybe @​_ is getting bound to the if's block rather than the subs block

From @smls

Still reproducible.

Simpler test-case, comparing `for` (which works as expected) to `if`​:

  for 42 { say $^a }; # 42
  for 42 { say @​_ }; # [42]

  if 42 { say $^a_ }; # 42
  if 42 { say @​_ }; # []

So it happens even without @​_ from an outer routine interfering.

Introspection reveals that the correct signature is created though​:

  for 42 { say &?BLOCK.signature; @​_ }; # (*@​_)
  if 42 { say &?BLOCK.signature; @​_ }; # (*@​_)

The `if` block just refuses to bind its argument to that signature, for some reason...

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

From @smls

Happens with an explicitly declared slurpy, too - regardless of its name​:

  for 42 -> *@​a { say @​a }; [42]
  if 42 -> *@​a { say @​a }; []

(This is Rakudo version 2016.04-210-gc59e4dc built on MoarVM version 2016.05
implementing Perl 6.c.)

From @dogbert17

On Fri, 20 May 2016 14​:34​:25 -0700, smls75@​gmail.com wrote​:

Happens with an explicitly declared slurpy, too - regardless of its name​:

for 42 \-> \*@​a \{ say @​a \};  \[42\]
if  42 \-> \*@​a \{ say @​a \};  \[\]

---
(This is Rakudo version 2016.04-210-gc59e4dc built on MoarVM version 2016.05
implementing Perl 6.c.)

Fixed by Zoffix++​:

Fix​: Raku/nqp@8e7201eb64
rakudo/rakudo@dfb6d951d7
rakudo/rakudo@ef1d22f4c1

Test​: Raku/roast@aa90737df1

@dogbert17 - Status changed from 'open' to 'resolved'