Code review comment for lp:~barry/lazr.config/lp1096512

On Jan 10, 2013, at 07:17 PM, Curtis Hovey wrote:

> 1. decode('ascii', 'strict'): appear to mean that someone upgrading might
> see more unicode errors raised.

Interestingly, the documentation says that only ASCII is acceptable but the
use of 'ignore' didn't enforce that, at least AFAICT.

> 2. sorted category names: is nice to have, but different from the past.

True. I thought about sorting in the test instead (you have to sort one place
or the other for doctest reproducibility). I mildly preferred to guarantee it
in the API.

Thanks for the review. I'll make sure the NEWS file and the release
announcement properly document these issues.