From mark@kli.org

 
 
 
 
  > (all("a", "b") ~~ /a/).Bool

  True

  > all("a", "b") ~~ /a/

  ｢all("a", "b")｣

  > (all("a", "b") ~~ /a/).perl

  Match.new(orig => all("a", "b"), from => 0, to => 1, ast
  => Any, list => ().list, hash => EnumMap.new())

  >

  What's going on is that the failed match returns a Nil, which
  disappears when it is added to the elements of the junction, so we
  get a junction of one element, which is True.

From @timo

What's going on is that the failed match returns a Nil, which
disappears when
it is added to the elements of the junction, so we get a junction of
one
element, which is True.

That's not quite right; the problem stems from the list iteration that happens to extract the slurpy arguments to the constructor of the junction; If we build a second constructor for junctions that takes a full array rather than slurpies, we can get around this problem even before the full list re-work is done.

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

From @usev6

This works now​:

< bartolin> r​: say all("a", "b") ~~ /a/ # RT #​120992
<+camelia> rakudo-{moar,jvm} 5d83db​: OUTPUT«all(｢a｣, Nil)␤»

I added a test to S03-junctions/misc.t with commit Raku/roast@50c2c6c8d9

I'm closing this ticket as 'resolved'.

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