GeoIP returns RS for Serbia which is the correct new iso code

Since geoip is correct, the bug is in the test (which is a rosetta one).

Bjorn suggested in the internal launchpad@ list: "Using live data makes the tests quite fragile, as already shown. I'd suggest registering a new GeoIP utility for tests, which would replace self._gi with a libGeoIP-like object. That object would return what we want it to return, and not depend on external data."

That is a valid approach.

Another approach is that we have a test that tests all our expectations about using the external data in our test suite.

That way, when the geoip data changes, then our test suite breaks. However, it's very obvious exactly what expectation has been broken.

My partial fix, that address the immediate problem of the failing test landed in RF 6247. Should we unduplicate my bug from this one? There are two issues here: One is a failing test that is inconsequential to Launchpad users, and the GeoIP issue that does affect users.

Curtis: I beg to differ.

The only user-affecting issue reported in this bug is that something has become correct. That does not need fixing. The flip side of that coin is "the correction isn't in production yet," which is indeed a separate problem with a separate solution—but it's to upgrade production systems, and that was never a part of either of these bugs. It's being handled elsewhere.