It's best to think of performance in two contexts: implementation and semantics.
Performance problems that stem from implementation can be fixed, and can yield significant benefits. This work is continually ongoing in Ruby (see the recent GC improvements) and the proliferation of alternate implementations will benefit this effort significantly (see the speed improvements to JS resulting from competition between V8, Nitro, and IonMonkey). That said, these sorts of improvements probably won't appear on any roadmap, since they can happen without the knowledge of the language's users.
Performance issues arising from language semantics, on the other hand, are a very real concern for the roadmap going forward. Here, I'm thinking specifically about the refinements feature. Charles Nutter had an excellent breakdown of how this feature might have far-reaching impacts on performance here: http://redmine.ruby-lang.org/issues/4085#note-46 . Additionally, there may be avenues to improve performance by subtly altering the semantics of existing features (such as 'define_method'). Unfortunately, while these changes/features will be discussed as part of the roadmap, performance is typically only addressed in a bit of a side-long manner (i.e. if you want to know what the performance impact will be, you'll probably need to pay attention to the discussions on ruby-core).
All that said, I think you can rest easy knowing that, even though performance may not be a first-class concern for the Ruby 2.0 effort, it is something that the community is very cognizant of and will be paying attention to.
Performance problems that stem from implementation can be fixed, and can yield significant benefits. This work is continually ongoing in Ruby (see the recent GC improvements) and the proliferation of alternate implementations will benefit this effort significantly (see the speed improvements to JS resulting from competition between V8, Nitro, and IonMonkey). That said, these sorts of improvements probably won't appear on any roadmap, since they can happen without the knowledge of the language's users.
Performance issues arising from language semantics, on the other hand, are a very real concern for the roadmap going forward. Here, I'm thinking specifically about the refinements feature. Charles Nutter had an excellent breakdown of how this feature might have far-reaching impacts on performance here: http://redmine.ruby-lang.org/issues/4085#note-46 . Additionally, there may be avenues to improve performance by subtly altering the semantics of existing features (such as 'define_method'). Unfortunately, while these changes/features will be discussed as part of the roadmap, performance is typically only addressed in a bit of a side-long manner (i.e. if you want to know what the performance impact will be, you'll probably need to pay attention to the discussions on ruby-core).
All that said, I think you can rest easy knowing that, even though performance may not be a first-class concern for the Ruby 2.0 effort, it is something that the community is very cognizant of and will be paying attention to.