> But that still means there is only one actual trusted server implementation, right?
Nope? Dendrite works. It's just stuck in beta for now due to lack of manpower.
> I thought the point of dendrite was to be able to prove that MSCs were implementable independently before they became part of the spec. It seems like the actual spec is just whatever MSC synapse has implemented.
The spec is not "whatever MSC synapse has implemented". The spec is... the spec, and the compliance test suites (sytest, complement) which go with it. Currently we require one implementation of MSCs to prove their validity before landing in the spec - and that could be in synapse, dendrite, conduit or wherever.
Just because the core team only has $ to actively progress a single server implementation right now doesn't devalue the spec at all. If anything it makes 3rd party implementations like the Conduits even more important, in terms of showing that the spec isn't coupled to the Synapse implementation.
> Is the "One chat protocol to bridge them all" goal abandoned?
Right now the goal is "have a decentralised open chat protocol whose apps can outperform mainstream centralised products". Bridging is secondary (for now, although Beeper is obviously focused on it).
> > the core team has ended up focusing our energy on a smaller set of projects while making sure we get a small set of things really excellent
> I'd love to see that. I'd especially love to see it stabilize and simplify enough that we have a multitude of servers, clients and bridges that are stable and useful.
The two are mutually exclusive. We can't simultaneously focus on getting a small number of implementations excellent... while also working on a multitude of implementations. Instead, we can make sure that the implementations that the core team works on at Element are great, solve problems, simplify things and unlock everyone else to be able to build better ones too - which is precisely what we're doing.
Nope? Dendrite works. It's just stuck in beta for now due to lack of manpower.
> I thought the point of dendrite was to be able to prove that MSCs were implementable independently before they became part of the spec. It seems like the actual spec is just whatever MSC synapse has implemented.
The spec is not "whatever MSC synapse has implemented". The spec is... the spec, and the compliance test suites (sytest, complement) which go with it. Currently we require one implementation of MSCs to prove their validity before landing in the spec - and that could be in synapse, dendrite, conduit or wherever.
Just because the core team only has $ to actively progress a single server implementation right now doesn't devalue the spec at all. If anything it makes 3rd party implementations like the Conduits even more important, in terms of showing that the spec isn't coupled to the Synapse implementation.
> Is the "One chat protocol to bridge them all" goal abandoned?
Right now the goal is "have a decentralised open chat protocol whose apps can outperform mainstream centralised products". Bridging is secondary (for now, although Beeper is obviously focused on it).
> > the core team has ended up focusing our energy on a smaller set of projects while making sure we get a small set of things really excellent
> I'd love to see that. I'd especially love to see it stabilize and simplify enough that we have a multitude of servers, clients and bridges that are stable and useful.
The two are mutually exclusive. We can't simultaneously focus on getting a small number of implementations excellent... while also working on a multitude of implementations. Instead, we can make sure that the implementations that the core team works on at Element are great, solve problems, simplify things and unlock everyone else to be able to build better ones too - which is precisely what we're doing.