From @Util

I tried to translated Perl 5​:
  my %bases=('K' => $base, 'M' => $base**2, 'G'=>$base**3, 'T'=>$base**4);
into both of these in Perl 6​:
  my %bases = <K M G T> Z=> ( $base X** 1..* );
  my %bases = <K M G T> Z=> ( $base <<**<< 1..* );

The cross-op version works.
The hyper-op version gives incorrect output​:
  <Util> m​: my $base = 10; my %bases = <K M G T> Z=> ( $base <<**<< 1..* ); say %bases.perl
  <camelia> rakudo-moar c2a57e​: OUTPUT«{​:G(12), :K(10), :M(11), :T(13)}<>␤»

Removing the interactions of Z=> and Whatever, lead to this simplified bug​:

< Util> m​: say [ 10 <<**<< (1,2,3,4) ].perl;
<+camelia> rakudo-moar c2a57e​: OUTPUT«[10, 100, 1000, 10000]␤»
< Util> m​: say [ 10 <<**<< (1..4) ].perl;
<+camelia> rakudo-moar c2a57e​: OUTPUT«Cannot find method 'Range'␤ in block <unit> at /tmp/nk6UBshobl​:1␤␤»
< Util> ??? Bug?
< masak> Util​: looks like.
< lizmat> repeatable WAT's should be perlbugged :-)
< masak> Util​: please submit.
< Util> lizmat​: Noted. masak​: will do.
< Util> lizmat​: I confess that when I post 2 (working,then non-working) `m​:` on this channel, and ask `bug?`
< Util> I am also secretly asking "is this a *known* bug, that I would already know if I read the channel as much as I should" :^)
< Util> thanks all!
< lizmat> Util​: fwiw, it didn't look familiar to me

Note that the problem persists when `**` is changed to `*`.
The result does change when the range is wrapped in parens.

<Util> m​: say [ 10 <<*<< (1,2,3,4) ].perl;
<camelia> rakudo-moar c2a57e​: OUTPUT«[10, 20, 30, 40]␤»
<Util> m​: say [ 10 <<*<< 1..4 ].perl;
<camelia> rakudo-moar c2a57e​: OUTPUT«[]␤»
<Util> m​: say [ 10 <<*<< (1..4) ].perl;
<camelia> rakudo-moar c2a57e​: OUTPUT«Cannot find method 'Range'␤ in block <unit> at /tmp/NvWvYfUj6y​:1␤␤»

Finally, I do not see the problem when the "magic" pointy-side of the hyper is a range;
the problem only shows up on the non-magic side.
<Util> m​: say (1,2,3,4) >>*>> (2..3);
<camelia> rakudo-moar c2a57e​: OUTPUT«2 6 6 12␤»

--
Thank you,
Bruce Gray (Util of PerlMonks)

From @usev6

One thing to note is that infix​:<**> (and also the hypered version of it) has a tighter precedence than the range operator. Therefore we need parenthesis to get the expected result. (Using a range there works now).

$ perl6-m -e 'say 1 <<**<< (1 .. 4)'
(10 100 1000 10000)

$ perl6 -e 'say 10 <<**<< 1 .. 4'
10..4

$ perl6 -e 'say (10 <<**<< 1) .. 4'
10..4

The thighter precedence was also behind the "incorrect" output of the compound example​:

$ perl6 -e 'my $base = 10; my %bases = <K M G T> Z=> ( $base <<**<< 1..* ); say %bases'
G => 12, K => 10, M => 11, T => 13

... which is the same as

$ perl6 -e 'my $base = 10; my %bases = <K M G T> Z=> ( ($base <<**<< 1) .. * ); say %bases'
G => 12, K => 10, M => 11, T => 13

Using parenthesis around the infinite range fails with X​::HyperOp​::Infinite​:

$ perl6 -e 'my $base = 10; my %bases = <K M G T> Z=> ( $base <<**<< (1 .. *) ); say %bases'
List on right side of hyperop of infix​:<**> is known to be infinite
  in block <unit> at -e​:1

But using a finite range works​:

$ perl6 -e 'my $base = 10; my %bases = <K M G T> Z=> ( $base <<**<< (1 .. 4) ); say %bases'
G => 1000, K => 10, M => 100, T => 10000

I added three tests to S03-metaops/hyper.t with commit Raku/roast@0e4520aff9.

I'm closing this ticket as 'resolved'.

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