Marius Vollmer

> what exactly is this blocked on currently? The original(ish) commit https://github.com/cockpit-project/cockpit/pull/22381/commits/b1f313eed379d0a6cd78a0d785ad5d03c8b1fd08 looked good to me, as an imperfect but effective workaround. (I have not seen evidence that popper can...

> The original(ish) commit [b1f313e](https://github.com/cockpit-project/cockpit/commit/b1f313eed379d0a6cd78a0d785ad5d03c8b1fd08) looked good to me, This is the real original: https://github.com/cockpit-project/cockpit/commit/48e813f2326db3bfa8da0ab1710eaa08ee91a8c7, sorry.

> Though this won't work for all scenarios and people might get confused. The cockpit container will only ever connect to to 127.0.0.1 when logging in locally,. What other scenarios...

This seems to still be draft, right?

@Venefilyn, in case you haven't already, please have a look at `pkg/shell/manifests.ts`, which validates those parts of the manifests that are of interest to the Shell. (I.e., not the "bridges"...

I like this! We should validate our own manifests according to this schema in test/common/static-code.

> We should do the same here. I am afraid that's not enough. It is nice to not send two exactly equivalent AddMatch requests to the D-Bus broker, but we...

> Instead, we should unconditionally install a single slot (per connection) with sd_bus_match_signal I misunderstood what sd_bus_match_signal does, I thought it does not send a AddMatch message to the broker....

@martinpitt, I think the last two steps in the diagram about "start bridge, connect stdio" and "start websocket" are confusing/wrongish and not necessary to explain the login process. The show...

@martinpitt, as discussed on chat, let's remove cockpit-tls from the diagram (and description) entirely.