String#= is 3 times slower in rubinius than in mri 2.1.0.preview1
It is approx 40 percent of the problem with the bm_word_anagrams.rb benchmark
which is 3 times slower in rubinius than mri
s = "abcdefghijk"
s = "x"
rubinius style benchmark: https://gist.github.com/kwleland/7988659
Any reason you didn't turn it into a PR with the benchmark?
just wanted to make sure it was sufficiently worthy :) Am working on it.
I consider adding useful benchmarks to be always worthy :). And if it turns out not to be, that can always be discussed on the PR :).
PR #2831 added