From @masak

<lizmat> rn​: my %h= (​:a<1>); say %h.postcircumfix​:<{ }>("a", p => 0 )
<camelia> rakudo fe7049​: OUTPUT«"a" => "1"␤»
<camelia> ..niecza v24-45-gbf8d2ad​: OUTPUT«"a" => 1␤»
* lizmat thinks that is a bug, it should just be "1", not a pair
<masak> lizmat++ # finding all these nice corner cases
<masak> lizmat​: want me to submit the p => 0 one?
<lizmat> masak​: please
* masak submits lizmat's rakudobug

Usually, just accessing a hash yields "1" in the above case. The :p
parameter says "give me the pair", and so passing in :p (as a named
parameter or an adverb, same thing) should indeed give "'a' => '1'".
But here lizmat is passing the named parameter with the value 0 (one
of the falsy values in Perl 6), which should again turn it off. (Use
case​: passing in a variable which may be truthy or falsy.)

From @masak

<lizmat> actually, the :p(0) also applies to :delete and :exists
<lizmat> rn​: my %a; %a<a>=1; say %a<a>​:delete(0); say %a
<camelia> rakudo 2a04f2​: OUTPUT«1␤().hash␤»
<camelia> ..niecza v24-50-gba63d9a​: OUTPUT«1␤{}␤»
<jnthn> lizmat​: Correct.
* masak adds that to the ticket

@masak - Status changed from 'new' to 'open'

From @usev6

There have been some relevant changes since this ticket was created​:

1 postcircumfix​:<{ }> is a sub now.

2 There was a change to the design documents and according to S02 the adverbial forms :p, :kv, :k, :v "weed out non-existing entries if the adverb is true; if not, they leave them in, just as an ordinary slice would" (cmp. Raku/old-design-docs@fd94c28225).

Taking these changes into account, now the mentioned commands run as expected. First the :p(0) stuff​:

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:p'
"a" => 1

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:p(0)' ## same result since key "a" exists
"a" => 1

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:!p' ## same thing
"a" => 1

$ perl6 -e 'my %h; %h<a>=1; say %h<b>​:p' ## key "b" does not exist

$ perl6 -e 'my %h; %h<a>=1; say %h<b>​:p(0)' ## pair not weeded out
"b" => Any

$ perl6 -e 'my %h; %h<a>=1; say %h<b>​:!p' ## same thing
"b" => Any

The adverbs :delete and :exists also work as designed​:

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:delete; say %h'
1

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:delete(0); say %h'
1
"a" => 1

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:exists'
True

$ perl6 -e 'my %h; %h<a>=1; say %h<a>​:exists(0)'
False

For :p(0) I added tests to S32-hash/pairs.t with commit Raku/roast@a543b07ab4

For :delete(0) and :exists(0) there are passing tests in S32-hash/delete-adverb.t and S32-exists-adverb.t, respectively.

I'm closing this ticket now.

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