Bridging group membership with 3rd party group sources

[benparsons]

Documentation: https://docs.google.com/document/d/1Nyk3Jf9BF0T2jHbeOV4DltazY5a3eP2meovSnMKDsxU/edit#heading=h.aienm7wdvf4q
Author: @lukebarnard1
Date: 15/11/2017

[richvdh]

Is this abandoned, or is it related to the way that the IRC bridge does the +irc group? @Half-Shot, any ideas?

[Half-Shot]

I've not seen this before, but it looks like we use 'Option C' for applying a group to a userid namespace (@freenode_.* is bridged to +freenode:matrix.org). Synapse supports this. Notably, the bridges have no awareness of this because it's a registration file option for synapse.

This is something that needs to be specced for the appservice registration file

[richvdh]

ok; it sounds like this is a thing that doesn't even affect the Matrix protocol, then. Accordingly I'm not sure this even qualifies as an MSC?

[Half-Shot]

It's definitely a protocol thing, since the AS spec defines the format of the registration file. Whether it's an MSC thing though is up to @anoadragon453 I guess.

[anoadragon453]

If this would just constitute a change to reg files, then that is more of an implementation/Synapse-specific item then a protocol change, so if that's the case then I do not think it should be an MSC.

However, we should probably refactor the spec on specifying options for appservices, by saying there are any number of ways it can be done, here is how Synapse does it, and link to a doc in the Synapse repo.

In the long term, we may be able to eliminate registration files altogether, which would help standardize appservices across every homeserver. I believe including the registration file in the current spec was in an effort to maintain standardization across home servers until a cleaner solution is found.

[turt2live]

re: changing registration files. https://github.com/matrix-org/matrix-doc/issues/1309 tries to propose a way to have a standard API, although the options included in the request are relatively implementation specific.

[turt2live]

Closing as groups were effectively rejected from the spec. This MSC appears to need significant re-work for Spaces.