* use base to make merging with bmo easier

* Revert "Bug 1497042 - Enclose table names in CREATE queries"

This reverts commit 62d5637a.

* refactor Bugzilla::DB to not subclass DBI

* use DBIx::Connector to manage database connections

* Ensure we always call DBIx::Connector->dbh before any DBI method

The code didn't allow a way of doing this without a lot of work.

So I had to take the following approach:
The 'dbh' attribute is now a method that delegates to DBIx::Connector's dbh
method. Per the docs, ->dbh() "Returns the connection's database handle. It will
use a an existing handle if there is one, if the process has not been forked or
a new thread spawned, and if the database is pingable. Otherwise, it will
instantiate, cache, and return a new handle."

Then there is the matter of the 'handles' on dbh. I've used Package::Stash to
insert proxy methods into the class when it is loaded.

* Bug 1328659 - Add support for utf8=utf8mb4 (switches to dynamic/compressed row format, and changes charset to utf8mb4)

* add deps

* add prepare_cached to the list of delegated methods.

This was added in bug 340160

* improve the migration to compressed/dynamic rows, skip views

* add missing semicolon

* remove pre-utf8mb4 emoji support

* fix small issue

* add debugging fatal error

blame jeff

Bug 828344: "contains all of the words" no longer looks for all words within the same comment or flag
r=LpSolit, a=LpSolit

r=dkl a=LpSolit

r/a=LpSolit

https://bugzilla.mozilla.org/show_bug.cgi?id=574577

Bug 680131: Replace the MPL 1.1 license by the MPL 2.0 one in all files, and add it to files which miss one
r=kiko r=mkanat r=mrbball a=LpSolit

r/a=LpSolit

catch known-broken tests as it's supposed to.
r=mkanat, a=mkanat (module owner)

r=mkanat, a=mkanat (module owner)

r=mkanat, a=mkanat (module owner)

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

the boolean chart operators
r=mkanat, a=mkanat (module owner)

r=mkanat, a=mkanat (module owner)

for "Attachment data"
r=mkanat, a=mkanat (module owner)

r=mkanat, a=mkanat (module owner)

wasn't working properly.
r=mkanat, a=mkanat (module owner)

including "nowordssubstr" and "notequals".
r=mkanat, a=mkanat (module owner)

and "dependson" fields.
r=mkanat, a=mkanat (module owner)

field in Search.pm. (This also makes OR searches work against the bug_group
field.)
r=mkanat, a=mkanat

tagging system
r=mkanat, a=mkanat (module owner)

combination in Search.pm.
r=mkanat, a=mkanat (module owner)

https://bugzilla.mozilla.org/show_bug.cgi?id=620827

r/a=mkanat

instead of searching the time on individual comments.
r=mkanat, a=mkanat (module owner)

fixed certain work_time tests in xt/search.t.

r=gerv a=LpSolit

r=mkanat, a=mkanat (module owner)

(Just some missing "this test is OK" bits for longdescs.count injection tests.)

r=mkanat, a=mkanat (module owner)

and PostgreSQL
r=mkanat, a=mkanat (module owner)

r=mkanat, a=mkanat (module owner)

resolution field
r=mkanat, a=mkanat (module owner)

r=mkanat, a=mkanat (module owner)

query parameters
r=mkanat, a=mkanat (module owner)

tests, so that the Pg tests properly understand that they can't test
creation_ts changedafter or changedbefore, *ever*.

https://bugzilla.mozilla.org/show_bug.cgi?id=600496