solnic / coercible

Powerful, flexible and configurable coercion library. And nothing more.
MIT License
137 stars 16 forks source link

1.8.7 out #9

Closed DawidJanczak closed 10 years ago

DawidJanczak commented 11 years ago

Hi,

According to discussion under https://github.com/solnic/virtus/pull/168 I think it's about time to drop 1.8.7 (along with backports) from coercible too. Please let me know if this is something you want to do now and if there are any possible issues with this change. If it's fine I will try and remove 1.8.7 support from other projects too (namely adamantium and mutant) so that backports can be dropped.

warmwaffles commented 10 years ago

Anything holding this back from happening?

dkubb commented 10 years ago

@warmwaffles nope ;)

warmwaffles commented 10 years ago

HAH, thank you

dkubb commented 10 years ago

Hm it seems as if the build breaks with the latest jruby-head on travis. I will look at this tomorrow unless someone beats me to it.

DawidJanczak commented 10 years ago

Thanks for the merge. As for the spec it seems to be a JRuby issue. The tests fail without this PR being merged too and they work fine with JRuby from two weeks ago. I will try to pinpoint JRuby commit that changed the behaviour, we should know more then.

dkubb commented 10 years ago

@DawidJanczak thanks! I appreciate you looking into this... it happens from time to time when travis upgrades some version of ruby. Sometimes it turns out it's a bug in their implementation, or maybe it points to an invalid assumption we've made.

DawidJanczak commented 10 years ago

No worries. This seems to be the commit that, when reverted, causes the spec to pass: https://github.com/jruby/jruby/commit/17b3966986c5bb85282ccf659b144f2a92d78276 It seems to make sense to have this behaviour and it's consistent with MRI, but I can't see how it would cause a recursive array to appear in that test scenario.

DawidJanczak commented 10 years ago

OK I have managed to track it down to this: https://github.com/jruby/jruby/issues/1176 Not sure if there is any quickfix for that so I fear that one test will be failing until it's fixed in JRuby.