Replacing labels fails due to wrong URL parsing

[dLobatog]

Currently octokit.rb uses CGI.escape for parsing the label. This makes strings like Waiting on contributor look like Waiting+on+contributor. The Github API v3 will not recognize such strings, as the spec (rightly so) says spaces should be encoded as %20, not +.

This renders the bot unable to replace the labels, and I rely on them to know at first glance which Pull Requests need a review. We cannot go back to < 3.0 Octokit because other prprocessor features depend on it.

I'm thinking of a possible patch, perhaps using ERB::Util.url_encode but it's going to be tricky as before sending the API request, Octokit tries to clean up the URL using URI::Parser, and that would convert every %20 into %2520.

Any ideas to get some hotfix in prprocessor appreciated. Aside from the obvious of adding hyphens or dashes to the label names 😄