Open p6rt opened 6 years ago
On the JVM backend the coercion type returned from .keyof for a regular hash is not the same object as Str(Any):
$ ./perl6-j -e 'use nqp; say nqp::eqaddr(Str(Any), {"a"=>1}.keyof)' 0
$ ./perl6-m -e 'use nqp; say nqp::eqaddr(Str(Any), {"a"=>1}.keyof)' 1
That's why some tests in S09-typed-arrays/hashes.t and S32-hash/perl.t have a new failure mode after fixing RT #132673. The objects are compared with infix:\<===> from Any.pm and that uses nqp::eqaddr. [1]
$ ./perl6-j -Ilib -e 'use Test; is({"a" => 1}.keyof, Str(Any))' not ok 1 - # Failed test at -e line 1 # expected: (Str(Any)) # got: (Str(Any))
I'm opening this ticket in order to update the todo message of said tests.
[1] https://github.com/rakudo/rakudo/blob/14c2cdb314ee5bbda34354ef535d4909dcea845a/src/core/Any.pm#L488
Migrated from rt.perl.org#132694 (status was 'new')
Searchable as RT132694$