From @dwarring

The perl method on Rakudo/MoarVM currently outputs has entries in a random order​:

  % perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
  {"c" => 30, "a" => 10, "b" => 20}
  % perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
  {"a" => 10, "b" => 20, "c" => 30}

JVM otoh, does seem to order consistantly​:

  % perl6-j -e'say { :a(10), :b(20), :c(30) }.perl'
  {"a" => 10, "b" => 20, "c" => 30}
  % perl6-j -e'say { :a(10), :b(20), :c(30) }.perl'
  {"b" => 20, "c" => 30, "a" => 10}

S02 doesn't have an opinion on hash ordering - http://design.perl6.org/S02.html#The_.perl_method

I'm in the habit of using the .perl method as part of the normal development/debugging progress, but I find that the MoarVM behavior can be a bit taxing in practice.

Could the MoarVM .perl method also order hash entries?

From @dwarring

Correction​: both MoarVM and JVM are re-ordering hash entries. Changing the title of this ticket.
On Sun Mar 22 12​:49​:22 2015, david.warring wrote​:

The perl method on Rakudo/MoarVM currently outputs has entries in a
random order​:

% perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
{"c" => 30, "a" => 10, "b" => 20}
% perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
{"a" => 10, "b" => 20, "c" => 30}

JVM otoh, does seem to order consistantly​:

% perl6-j -e'say { :a(10), :b(20), :c(30) }.perl'
{"a" => 10, "b" => 20, "c" => 30}
% perl6-j -e'say { :a(10), :b(20), :c(30) }.perl'
{"b" => 20, "c" => 30, "a" => 10}

S02 doesn't have an opinion on hash ordering -
http://design.perl6.org/S02.html#The_.perl_method

I'm in the habit of using the .perl method as part of the normal
development/debugging progress, but I find that the MoarVM behavior
can be a bit taxing in practice.

Could the MoarVM .perl method also order hash entries?

From @pmichaud

On Sun Mar 22 13​:04​:21 2015, david.warring wrote​:

Could the MoarVM .perl method also order hash entries?

Requiring .perl to order hash entries ought to be a language/specification change, rather than something that individual compilers do on their own. Otherwise programmers may falsely rely on a compiler-specific behavior.

To have this considered as a language feature, add an issue to GitHub​: https://github.com/perl6/specs/issues

Pm

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

From @dwarring

On Mon Mar 23 06​:48​:38 2015, pmichaud wrote​:

On Sun Mar 22 13​:04​:21 2015, david.warring wrote​:

Could the MoarVM .perl method also order hash entries?

Requiring .perl to order hash entries ought to be a
language/specification change, rather than something that individual
compilers do on their own. Otherwise programmers may falsely rely on
a compiler-specific behavior.

To have this considered as a language feature, add an issue to GitHub​:
https://github.com/perl6/specs/issues

Pm

Have opened issue Raku/old-design-docs#88
- David

From @dwarring

The .perl method on Hashes has been reworked in the latest rakudo. Looks good! (Looks Good!!).

% perl6-m --version
This is perl6 version 2015.04-204-ga040b1a built on MoarVM version 2015.04-62-g052aca0
% perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
{​:a(10), :b(20), :c(30)}
%
% perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
{​:a(10), :b(20), :c(30)}

Hashes currently appear to be ordered.

On Mon Mar 23 11​:09​:33 2015, david.warring wrote​:

On Mon Mar 23 06​:48​:38 2015, pmichaud wrote​:

On Sun Mar 22 13​:04​:21 2015, david.warring wrote​:

Could the MoarVM .perl method also order hash entries?

Requiring .perl to order hash entries ought to be a
language/specification change, rather than something that individual
compilers do on their own. Otherwise programmers may falsely rely on
a compiler-specific behavior.

To have this considered as a language feature, add an issue to GitHub​:
https://github.com/perl6/specs/issues

Pm

Have opened issue Raku/old-design-docs#88
- David

From @coke

On Sun May 10 13​:20​:24 2015, david.warring wrote​:

The .perl method on Hashes has been reworked in the latest rakudo.
Looks good! (Looks Good!!).

% perl6-m --version
This is perl6 version 2015.04-204-ga040b1a built on MoarVM version
2015.04-62-g052aca0
% perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
{​:a(10), :b(20), :c(30)}
%
% perl6-m -e'say { :a(10), :b(20), :c(30) }.perl'
{​:a(10), :b(20), :c(30)}

Hashes currently appear to be ordered.

On Mon Mar 23 11​:09​:33 2015, david.warring wrote​:

On Mon Mar 23 06​:48​:38 2015, pmichaud wrote​:

On Sun Mar 22 13​:04​:21 2015, david.warring wrote​:

Could the MoarVM .perl method also order hash entries?

Requiring .perl to order hash entries ought to be a
language/specification change, rather than something that
individual
compilers do on their own. Otherwise programmers may falsely rely
on
a compiler-specific behavior.

To have this considered as a language feature, add an issue to
GitHub​:
https://github.com/perl6/specs/issues

Pm

Have opened issue Raku/old-design-docs#88
- David

.perl on hashes is now sorted - all the tests in S32-hash/perl.t implicitly cover this condition.

Resolving ticket.
--
Will "Coke" Coleda

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