This continues the hack that cane assumes UTF-8 input, which may not actually be the case but still seems like a decent option since I don't have good examples of this not being the case. This should not add any performance overhead, since input is only re-encoded if any comparison fails on it. I have not benchmarked whether this actually a big deal or not, but should. Fixes #22.
The exact spec for this comment is unclear from the Ruby source. There is exactly one test for a valid comment, and `parse.y` is ... obtuse. The regex used here was found in `lib/erb.rb` and copied verbatim, with the exception of a redundant regex to exclude -*- markers, which AFAICT is entirely unnecessary. Fixes #21.
Previously, the doc check would extract a class name like "Foo;" from "class Foo; end"
* When adding cane to a legacy project and ratching down the max_violations setting it's handy to know what value to set * At the moment I need to manually sum the violation count for each grouping
…the parts of cane that use the existing tailor. Thought I'd try to preempt that here.