#10 06-seed.t failing test

[Anonymous]

The 10th test at 06-seed.t is failing randomly, both in Windows (Perl 5.22) and Linux (Perl 5.18). The test compares arrays (is_deeply \@dataset1, \@dataset2;) that ultimately come from the same seed; I have not checked the code carefully, but my guess is that somewhere in the middle there maybe hashes returning keys in random order (current default 'keys' behaviour) which are making both arrays different.

Some examples below:
/////
Failed test at t/06-seed.t line 51.
Structures begin differing at:
$got->[0]{end} = '262'
$expected->[0]{end} = '203'
Looks like you failed 1 test of 10.
/////
$got->[2]{end} = '227'
$expected->[2]{end} = '155'
////
$got->[1]{seq} = 'ccccccccccaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaatttttttttt'
$expected->[1]{seq} = 'tttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt'
////

Cheers,

Francisco J. Ossandon

[floflooo]

Seems to work fine on my side with perl 5.20.2

Could you try to individually upgrade each perl module that Grinder depends on with 'sudo cpan ...' to see if this improves things? Thanks,

Florent

[Anonymous]

Hello Florent,
I just realized that at the moment of generating this report my copy was not totally up to date. It only had commits until 2014-01-14!!! So the hash randomization fixes that you added later were not included. Now that I updated my copy it worked fine.

This was totally my fault, sorry about that. =(

Please close this report.

[Anonymous]

Ok, now I discovered something else... Before writing this report, I created a new folder with the updated repo, so the copy was indeed updated, but when I run the test from that copy, Perl went to my original outdated folder because it was in PATH!! That's why it gave me the wrong results.

So my fault was not about updating the repo but not realizing that because of PATH Perl was not using the updated folder. XD

Sorry anyway. :)

[Francisco J. Ossandon]

Please close this ticket. ;)

[floflooo]

Ok, good, closing.