Prior behavior was to show fallbacks as additions, which confused the difference between old and new. For the locale_changes, the trick is to reduce the milestone_signoffs query to mappings where both the signoff and the milestone are for the current appversion. For about-milestone, we need to get the fallbacks listed, and exclude them from the added data. For a better overview of what's new and what's old, I'm showing them in an extra column.
Trying to do months exactly didn't work out well, doing just 30.5 days intervals instead, and make the code logic way more consistent. Also, the checks for date-based queries need to work for just start or end dates.
Also, extend the ldap search pattern by 'mail=%s,o=net,dc=mozilla' as that seems to be used for volunteers now. Removing o=net,dc=mozillacom, that isn't used these days.
The original command is in bug 1152747.
…elmer We keep 7 days of build data, and 1 day of log files. Also keep the data for the latest build for each builder, as buildbot uses that for startup.
Another regression from the elmo package, bug 934547. The csrf monkeypatch creates a new django.contrib.admin.site object, so we need to run the autodiscovery after that happened.
Adding a dummy test case wrapper in elmo, right now this just does what test_utils.TestCase does. More importantly, though, drop using the RadicalTestSuiteRunner, as that's bringing problems for django versions 1.6 and up. This commit is mimicking what sumo does in bug 971014 and mozilla/kitsune@a8684ec
…, rs=foopy Rewrite the command to not depend on pyquery, also remove the dependency. The code also didn't know about the locale-team mappings, that's added now. Also moves from using print to self.stdout, like django commands should. I also took the liberty to rewrite the logic until I understood what it does again. It now generates a proposed VERBATIM_CONVERSIONS mapping as it goes, and compares that to the reference.