TST: don't fail on systems with tzdata backzone enabled

[rebecca-palmer]

test_dst.py specifies Africa/Kinshasa but expects a transition date (AmbiguousTimeError) that is actually for Africa/Lagos. On systems that enable tzdata's backzone file (which Debian and Ubuntu unstable now do), these are different, and the test hence fails (log).

This makes the obvious fix.

[rebecca-palmer]

The Windows timeout seems to be something that "randomly" happens (see other recent failures). Maybe we should increase the timeout, or if that's too expensive, run only a subset of tests there?

[MarcoGorelli]

Thanks @rebecca-palmer

No objections to merging, would just appreciate it if you explain the issue a little more

Is it that Africa/Kinshasa is erroneously raising AmbiguousTimeError for 1905-07-01, and that it shouldn't be doing so, so you're replacing it with Africa/Lagos, which correctly raises AmbiguousTimeError for that date?

thanks!

[github-actions]

This pull request is stale because it has been open for thirty days with no activity. Please update and respond to this comment if you're still interested in working on this.

[MarcoGorelli]

gentle ping (would much appreciate a little explanation of why this is necessary, I couldn't follow it from the link)

[rebecca-palmer]

Africa/Lagos has a transition, and hence the AmbiguousTimeError the test checks for, on that date.

On systems without backzone enabled, Africa/Kinshasa is a copy of Africa/Lagos (because they're the same timezone since before 1970(?)), and also does that.

However, on systems with backzone enabled, Africa/Kinshasa has its own timezone history, which doesn't have a transition on that date. This causes the test to fail on such systems.

This fixes this by instead using Africa/Lagos in the test.

[MarcoGorelli]

Makes sense - thanks @rebecca-palmer !