Bug #109435 for Tao-DBI: Random test failures with perl 5.17.x and later (hash randomization?)

Sun Nov 22 13:51:07 2015 SREZIC [...] cpan.org - Ticket created

Subject:

Random test failures with perl 5.17.x and later (hash randomization?)

See http://matrix.cpantesters.org/?dist=Tao-DBI%200.0008 for a fail/pass overview --- the pattern (fail reports start with perl 5.17.6) looks like there's a hash randomization issue here. Hint: if you are using Data::Dumper, then you should use the Sortkeys option for predictable hash key order.

Sun Jan 03 19:51:32 2016 ferreira [...] shoo.cpan.org - Correspondence added

The failures in t/006deep.t are due to a forgotten dependency of the test on the YAML module. As it seems, most of the smokers would have this module, which explains the many successful runs of the test. I am preparing 0.01 release with a fix for this. Thanks! On Sun Nov 22 13:51:07 2015, SREZIC wrote: Show quoted text

> See http://matrix.cpantesters.org/?dist=Tao-DBI%200.0008 for a > fail/pass overview --- the pattern (fail reports start with perl > 5.17.6) looks like there's a hash randomization issue here. > > Hint: if you are using Data::Dumper, then you should use the Sortkeys > option for predictable hash key order.

Sun Jan 03 19:51:33 2016 The RT System itself - Status changed from 'new' to 'open'

Sun Jan 03 20:19:21 2016 ferreira [...] shoo.cpan.org - Taken

Mon Jan 04 01:22:38 2016 SREZIC [...] cpan.org - Correspondence added

Well, the 0.01 shows the same random fail/pass pattern. See http://fast-matrix.cpantesters.org/?dist=Tao-DBI%200.01 On 2016-01-03 19:51:32, FERREIRA wrote: Show quoted text

> The failures in t/006deep.t are due to a forgotten dependency of the > test on the YAML module. As it seems, most of the smokers would have > this module, which explains the many successful runs of the test. I am > preparing 0.01 release with a fix for this. Thanks! > > On Sun Nov 22 13:51:07 2015, SREZIC wrote:

> > See http://matrix.cpantesters.org/?dist=Tao-DBI%200.0008 for a > > fail/pass overview --- the pattern (fail reports start with perl > > 5.17.6) looks like there's a hash randomization issue here. > > > > Hint: if you are using Data::Dumper, then you should use the Sortkeys > > option for predictable hash key order.

Mon Jan 04 01:22:47 2016 SREZIC [...] cpan.org - Broken in 0.01 added

Tue Jan 05 23:53:21 2016 ferreira [...] shoo.cpan.org - Correspondence added

You are completely right. I don't know why when I sample a few test failures – all I saw was the trouble with the undeclared YAML dependency. But the failures due to hash randomization were quite clear now. Release 0.011 should clear this for good. Thanks! On Mon Jan 04 01:22:38 2016, SREZIC wrote: Show quoted text

> Well, the 0.01 shows the same random fail/pass pattern. See > http://fast-matrix.cpantesters.org/?dist=Tao-DBI%200.01

Wed Jan 06 01:32:16 2016 SREZIC [...] cpan.org - Correspondence added

There are still random failures in 0.011. It seems you misspelled Sortkeys (no CamelCase here!). This patch should help: diff --git i/lib/Tao/DBI/st_deep.pm w/lib/Tao/DBI/st_deep.pm index d7224d1..43a2174 100644 --- i/lib/Tao/DBI/st_deep.pm +++ w/lib/Tao/DBI/st_deep.pm @@ -31,7 +31,7 @@ sub initialize { sub to_perl { require Data::Dumper; - local $Data::Dumper::SortKeys = 1; + local $Data::Dumper::Sortkeys = 1; return Data::Dumper::Dumper(shift); } diff --git i/t/005tr_hash.t w/t/005tr_hash.t index ab17aed..38b3f42 100644 --- i/t/005tr_hash.t +++ w/t/005tr_hash.t @@ -5,7 +5,7 @@ use_ok('Tao::DBI::st_deep'); sub to_perl { require Data::Dumper; - local $Data::Dumper::SortKeys = 1; + local $Data::Dumper::Sortkeys = 1; return Data::Dumper::Dumper(shift); } On 2016-01-05 23:53:21, FERREIRA wrote: Show quoted text

> You are completely right. I don't know why when I sample a few test > failures – all I saw was the trouble with the undeclared YAML > dependency. But the failures due to hash randomization were quite > clear now. Release 0.011 should clear this for good. Thanks! > > On Mon Jan 04 01:22:38 2016, SREZIC wrote:

> > Well, the 0.01 shows the same random fail/pass pattern. See > > http://fast-matrix.cpantesters.org/?dist=Tao-DBI%200.01

Fri Jan 08 01:55:49 2016 ferreira [...] shoo.cpan.org - Status changed from 'open' to 'resolved'

Fri Jan 08 01:55:49 2016 ferreira [...] shoo.cpan.org - Fixed in 0.012 added