Grammar for room IDs (SPEC-389)

[matrixbot]

Room IDs (and Event IDs, prior to room v3) are opaque to the user (they should never be seen), but they do have some syntax within the protocol; this means that they can be more baroque than User IDs, but are not as freeform as some other IDs (https://github.com/matrix-org/matrix-doc/issues/666).

We should ensure their syntax is well-specified, and make sure we know which characters they can include.

(Imported from https://matrix.org/jira/browse/SPEC-389)

(Reported by @richvdh)

[matrixbot]

Jira watchers: @richvdh

[matrixbot]

Links exported from Jira:

relates to SPEC-1
relates to matrix-org/matrix-spec-proposals#668

[matrixbot]

From [~markjh]:

... the v2 filter API makes assumptions that '*" is not a valid character in a room id or a user ID so that it can use it as a wildcard.

-- @richvdh

[matrixbot]

The local-part could have the same grammar as the completely-opaque IDs (https://github.com/matrix-org/matrix-doc/issues/666)?

-- @richvdh

[matrixbot]

Probably also relevant to https://github.com/matrix-org/matrix-doc/issues/666, the character set should probably be limited to avoid things that would be ambiguous or "annoying" when displayed, like whitespace and newlines. Also, if the opaque ID can contain a colon, that would make it harder to figure out which colon delimits the opaque ID from the IP/hostname and from the port.

-- Jimmy Cuadra

[richvdh]

As reflected in matrix-org/synapse#8409, allowing / in room ids causes issues with apache rejecting %2F.