Hash.clone differs from hash assignment; the hash and its clone are spookily entangled

[p6rt]

Migrated from rt.perl.org#127704 (status was 'resolved')

Searchable as RT127704$

[p6rt]

From @masak

<masak> m​: my %h1; %h1<foo> = 1; my %h2 = %h1; %h2<bar> = 2; say %h1
<camelia> rakudo-moar 42fb81​: OUTPUT«foo => 1␤»
<masak> m​: my %h1; %h1<foo> = 1; my %h2 := %h1.clone; %h2<bar> = 2; say %h1
<camelia> rakudo-moar 42fb81​: OUTPUT«bar => 2, foo => 1␤»
* masak submits rakudobug
<masak> pretty sure this one is a regression. 007 has had code based
on Hash.clone for a long while to indicate "and now we enter a new
scope"
<masak> the bug manifests in the fact that old variables don't go away
when we leave a scope
<moritz> m​: my %h1; %h1<foo> = 1; my %h2 = %h1.clone; %h2<bar> = 2; say %h1
<camelia> rakudo-moar 42fb81​: OUTPUT«foo => 1␤»
<moritz> masak​: I see
<moritz> the fact that one often uses hash assignment (which copies)
seems to shadow this bug from every day observation
<masak> moritz​: reason I didn't simply use hash assignment is that in
my case, the hash was in an array.
<moritz> star-m​: my %h1; %h1<foo> = 1; my %h2 := %h1.clone; %h2<bar> =
2; say %h1
<camelia> star-m 2016.01​: OUTPUT«bar => 2, foo => 1␤»
* moritz wonders if a simple method clone in class Hash or Map helps
<moritz> m​: say Hash.^mro
<camelia> rakudo-moar 42fb81​: OUTPUT«((Hash) (Map) (Cool) (Any) (Mu))␤»

[p6rt]

From @moritz

On Mon Mar 14 01​:40​:19 2016, masak wrote​:

<masak> m​: my %h1; %h1<foo> = 1; my %h2 = %h1; %h2<bar> = 2; say %h1
<camelia> rakudo-moar 42fb81​: OUTPUT«foo => 1␤»
<masak> m​: my %h1; %h1<foo> = 1; my %h2 := %h1.clone; %h2<bar> = 2; say %h1
<camelia> rakudo-moar 42fb81​: OUTPUT«bar => 2, foo => 1␤»
* masak submits rakudobug

Fixed in rakudo/rakudo@6b2b34a791, tested in Raku/roast@ab74762da1

Cheers,
Moritz

[p6rt]

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

[p6rt]

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