About cluster/scalability

[rom1504]

It could be interesting in the future to be able to scale flying-squid between several node instances (~ https://nodejs.org/api/cluster.html ) and several servers.
In order to do that we'd have to implement a good api to make the nodes have the same data.
One way would be putting all the data in prismarine-world and either:

1. doing a thing to keep prismarine-world in memory and in sync in all instances : not great
2. run prismarine-world in a single place and make some kind of network api to access it
3. make a prismarine-world-<some database> that would just communicate with a database and not store data in memory

Option 2 and 3 are mostly the same, with a different kind of implementation.

(anyway, this is completely long term and should not really be in "priority" there https://github.com/mhsjlw/flying-squid/blob/master/TODO.md#priority )

[rom1504]

This issues is related with data saving/loading though.
When deciding on the saving/loading architecture/api, we might be able to make this easier too.

[109C]

For number two, would the communication be world <--> worker <--> client, or worker <--> world <--> client?

[rom1504]

it would be worker <--> prismarine-world instance

[109C]

So the world instance would communicate with the client?

[rom1504]

the world instance would communicate with a worker yes, that worker would communicate (and handle) with minecraft clients
when needed that worker would somehow (through a master or directly) communicate with other workers to communicate with other mc clients

something like that.

but the most difficult part is taking the data out of the worker imho ;) (and into the world instance)

[BasToTheMax]

How would things like redstone be handled?