This is a big job. Here's what we know.

JRuby 1.7 and 9k require logic looks mostly identical, so I don't think this is a bug in require. Most likely this is a bug that has always existed in how we search for loaded and loading features, and it is being exposed now because 9k is (properly, I think) expanding absolute paths at various places during load.

I started looking into MRI's logic, and it has many differences from ours.

• search_required, which should be equivalent to our featureAlreadyLoaded, has a great deal more logic. It is extension-aware, and performs lower level searches for loaded features with the extension in mind.
• rb_feature_p, upon which we probably based our featureAlreadyLoaded, has logic to check expanded load path entries, circular load lock ("loading" collection) and more.
• MRI's feature_index appears to aggregate not just the file loaded for a feature, but other file paths that should match that feature.

I'm attempting a partial port, but I have other responsibilities that will pull me away from this.

I started the port of search_required. Here's what I have: https://gist.github.com/headius/a78c68145e2f8c739b1e

This is mostly finished except for the goto in there.

The big snag came while attempting to port rb_feature_p, due to differences in our feature_index and how we maintain it.

We also don't have an expanded load_path as in MRI. They use this expanded load_path cache for trying different filesystem prefixes.

On the up side, if we manage to port all this logic over it could speed up require/load searching too.

I will be unable to work on this for a while. Pinging @ratnikov and @Who828 since they might have interest in picking up where I left off.

We actually had to fix this to get Rails working, because an autoloaded class that undef'ed a method was getting run twice (and blowing up). Should be fixed in 9k.pre2.

Apparently a big problem for others on 1.7, so I backported the fixes there too.