raw SQL.
r=mkanat, a=mkanat (module owner)

the changedbefore/after charts include the date specified
(they previously did exclusive searches)
r=mkanat, a=mkanat (module owner)

(and a few as being not broken at all on Pg).
r=mkanat, a=mkanat (module owner)

so I changed the values to make them consistent. (I suspect the inconsistency
is a bug in Pg itself that is not actually important to normal Bugzilla
usage.)

in a loop
r=glob, a=mkanat

<> values into the actual bug value, for searching
r=mkanat, a=mkanat (module owner)

because they make Pg throw an error and then be unable to run any further
tests. It's OK to skip these tests because they still run on MySQL, so
we'll still catch any injection vulns.
r=mkanat, a=mkanat (module owner)

https://bugzilla.mozilla.org/show_bug.cgi?id=575021

Instead, we pass around a hashref and update the hashref. This patch also
includes some cleanup for bugs surrounding percentage_complete,
attachments.isobsolete, attachments.ispatch, and owner_idle_time.
r=mkanat, a=mkanat

https://bugzilla.mozilla.org/show_bug.cgi?id=562014

incorrectly.
r=mkanat, a=mkanat (module owner)

boolean charts
r=glob, a=mkanat