bugs test parameter is outdated and Synapse-specific

[anoadragon453]

Dendrite relies on having a whitelist of tests that it can say it's currently passing. Each time a PR allows more tests to pass, the PR-ee has to add more tests to the testfile. We have a script which will tell the user to do this if a test is passing that we expected to fail.

There's some functionality in Sytest that allows you to put a bug parameter into a test, seen in a few places, but notably

sytest/tests/61push/07_set_enabled.pl

Lines 60 to 63 in 0d4ca96

	test "Enabling an unknown default rule fails with 404",
	requires => [ local_user_fixture() ],
	bug => "SYN-676",

and

sytest/tests/41end-to-end-keys/01-upload-key.pl

Lines 38 to 41 in fb631ec

	test "Should reject keys claiming to belong to a different user",
	requires => [ $fixture ],
	bug => "synapse#1396",

Dendrite currently passes these tests, but the CI complains that we should be marking them as whitelisted. They are being marked as whitelisted though. While whitelisting should prevent a test from being marked as expected fail, the bug parameter still does so. The code for that is here:

sytest/run-tests.pl

Line 588 in 24678f5

$params{expect_fail}++ if $params{bug} and not $FIXED_BUGS{ $params{bug} };

These bugs are specific to Synapse, and should probably be replaced with a // TODO (Synapse): instead. Alternatively, we could create a bugs parameter that's specific to each homeserver, but I feel that's a decent chunk of work and ups the barrier of entry of adding support for more homeservers.

[anoadragon453]

@richvdh and I have come up with a plan to remove the bug parameter and instead replicate the functionality using test blacklists in Synapse.

[anoadragon453]

This is fixed by:

matrix-org/synapse#5611
#638

[Cnly]

The PR didn't close this issue though...

[anoadragon453]

Indeed, oddly.

[richvdh]

(it won't be closed until the PR hits the default branch (master))