Fixed a small, yet major typo.

[Valarissa]

:dependant => :destroy is not a valid association parameter. :dependent => :destroy, however, is legit. This may have left us with a huge number of orphaned authorizations.

The authorizations on rstat.us proper should probably be checked to see if there have been any orphans left from account deletions.

Furthermore, it may actually be affecting the random test failures on travis... maybe.

[steveklabnik]

Oh man, did I do this?

[Valarissa]

46c227a » steveklabnik
2011-03-13 let's syndicate to twitter, shall we? 70 many :authorizations, :dependant => :destroy

Your name... and commit says yes, that was you. :)

[travisbot]

This pull request passes (merged a613450 into a3f66a5).

[steveklabnik]

fffuuuuuu

[jaimeiniesta]

That's why I think it's a good idea to test cascade deletions, validations, etc.

You're not testing that Rails knows how to do this, you're testing that you set this up properly in your models.

[carols10cents]

I still have no idea what causes the random travis failures that are skips right now... if it was this, shouldn't they fail locally too? Those tests never fail locally for me.

[Valarissa]

How many people here use Linux? I know we encountered something at Pikimal where our ci servers would run tests in a different order than Mac OSX due to differences in the file system crawling.

[carols10cents]

not it. minitest runs tests in random order by default though.... but i dont know what it does with nested describes? sounds plausible, i'm just not sure how likely.

[Valarissa]

I'm seeing the passed tests at the same point every run... which is odd because they definitely have different seeds.