test with predictable hashes

[szabgab]

No description provided.

[mschilli]

Wouldn't this skip all tests on older perls that don't have this hash jumble feature at all and therefore don't cause the problem?

[szabgab]

Oh right, so you probably also need to condition it on >= 5.18

[szabgab]

closed by mistake (screen is to small and I only saw the close button :)

[mschilli]

Can you explain what the code does? On my Macbook perl neither of these variables are set. Are they set on yours?

[szabgab]

Hmm, what I did there was I think a bit stupid. Only the problematic cases should be wrapped in the skip block and that should also be conditioned on the version of Perl. Then you'd run the test as

PERL_PERTURB_KEYS=0 PERL_HASH_SEED=1 prove -lv t/001Basic.t

or as

PERL_PERTURB_KEYS=0 PERL_HASH_SEED=1 make test

[mschilli]

Does adding

$ENV{ PERL_PERTURB_KEYS } = "DETERMINISTIC";
$ENV{ PERL_HASH_SEED } = 1;

to t/001Basic.t also work with your perl version?

[szabgab]

As far as I understand (and checked) one needs to set those environment variable before perl starts to run, so we cannot do it in the test script. It has to be done in the environment. See the examples in the code.
I think the DETERMINISTIC (2) will not work properly as that would mean the results are different on older perls and newer perls. from perldoc perlrun:

 PERL_PERTURB_KEYS
               (Since Perl 5.18.0)  Set to "0" or "NO" then traversing keys will be repeatable
               from run to run for the same PERL_HASH_SEED.  Insertion into a hash will not
               change the order, except to provide for more space in the hash. When combined
               with setting PERL_HASH_SEED this mode is as close to pre 5.18 behavior as you can
               get.

but I have not updated the comment yet.

I think the most recent change I pushed will work. I tested it on 5.18.0 and now it works both with and without the environment variables.