From dave.whipp@gmail.com

It's seems that Rakudo's notion of equality under the "===" operator doesn't
handle junctions​:

<dwhipp> pugs​: say "yes" if 1 === 1|2
<p6eval> pugs​: OUTPUT«yes␤»
<dwhipp> rakudo​: say "yes" if 1 === 1|2
<p6eval> rakudo 34350​: RESULT«Bool​::False»

From @pmichaud

On Thu Dec 25 09​:36​:49 2008, dave.whipp@​gmail.com wrote​:

It's seems that Rakudo's notion of equality under the "===" operator
doesn't
handle junctions​:

<dwhipp> pugs​: say "yes" if 1 === 1|2
<p6eval> pugs​: OUTPUT«yes␤»
<dwhipp> rakudo​: say "yes" if 1 === 1|2
<p6eval> rakudo 34350​: RESULT«Bool​::False»

Now fixed in r34390, thanks!

(Note​: Jonathan is working on a new dispatcher for Rakudo that will
handle junction autothreading -- until then we have to make special MMD
entries for any operators or functions that we want to be "junction
aware". See build/gen_junctions_pir.pl.)

Pm

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

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

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

From @pmichaud

Reopening ticket (and assigning to moritz++) -- we probably need a test
for this somewhere before closing the ticket.

Thanks,

Pm

From @pmichaud

I had to revert the change that autothreaded infix​:<===>, as the
existing non-autothreaded version was being used to create Junctions
(and resulted in an infinite loop).

I suppose it's also a legitimate question as to whether infix​:<===>
should autothread at all -- i.e., for something like​:

  my $a = 1;
  my $b = any(1, 2, 3);

  say 'yes' if $a === $b;

there might be an argument that infix​:<===> should return false since $a
and $b have different types (one is an Int, the other a Junction). We
might need to check with p6l on this.
Pm

From @pmichaud

On Fri Dec 26 10​:34​:01 2008, pmichaud wrote​:

I suppose it's also a legitimate question as to whether infix​:<===>
should autothread at all [...]

18​:56 <TimToady> pmichaud​: an exact === can always be emulated by
comparing .WHICH, and I think .WHICH (being a macro) can probably return
the junction's id rather than a
  superposition of ids
18​:56 <TimToady> so === should probably autothread, but junctional logic
shouldn't be written in terms of it, except where recursion is desirable
18​:59 <pmichaud> TimToady​: makes sense, thanks.
18​:59 <pmichaud> the place where infix​:<===> was being used was to
collapse duplicate values in any,all,none junctions.
18​:59 <pmichaud> (that can of course be fixed.)

Pm

From @moritz

Now tested in t/spec/S03-operators/value_equivalence.t.

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