On Wed, Feb 22, 2017 at 10:04 AM, Robert Bradshaw < rober...@google.com.invalid> wrote:
> with their respective costs (e.g. runtime, > dependencies) expected utility (how likely it is to find something) and > actionability (how easy it would be for someone to fix the reported errors > without asking us). > I'd add "detriment to newcomer morale / likelihood to deter new contributors", though obviously not really quantifiable we could say "none, little, some, lots". This isn't redundant with other criteria, as checkstyle is pretty fast, has almost no false positives, has actionable messages, and errors are easy to fix, but it is also nitpicky and a needless deterrent. Or, if you mean "how likely it is to find a real error" the score is zero. I am 100% in favor of checkstyle on the committed code, just not for the uninitiated. But yes, this is just my vote and rationale. I will certainly be happy with the consensus of the community. I agree that a multifaceted survey like Robert suggests might be helpful.
