Split controller OSC off from SuperDirt handshake OSC #1027
Conversation
|
Pinging @Zalastax, who's expressed interest in OSC listener/stream logic, but happy to hear from anyone else with thoughts as well |
|
|
||
| resolve :: String -> String -> IO N.AddrInfo | ||
| resolve :: String -> Int -> IO N.AddrInfo |
There was a problem hiding this comment.
Would it be a good idea to introduce a type synonyme here?
type Port = Int
| then Just <$> resolve (oAddress target) (show $ fromJust $ oBusPort target) | ||
| else return Nothing | ||
| cxs <- mapM (\(target, os) -> do remote_addr <- resolve (oAddress target) (oPort target) | ||
| remote_bus_addr <- sequence (resolve (oAddress target) <$> (oBusPort target)) |
There was a problem hiding this comment.
Smart! This uses the fact that Maybe is a Monad so sequence is the same thing as what was done here before.
|
Sorry for being a bit slow, somehow missed this, but looks great to me! |
|
@mindofmatthew which of the 4 points you mentioned in #982 (comment) does this address? I'm guessing 1-2? Good stuff! |
|
While trying to verify that this works correctly, I discovered that Tidal doesn't work correctly (see #1030), so I'm going to try and patch that here as well. @jarmitage Yes, this basically cleans up the Stream module a bit. It doesn't totally separate the listener (which is my eventual plan), but it splits up two listener functions (external controllers vs. getting info from SuperDirt) that are currently combined. |
|
@mindofmatthew are you happy to merge? (I'm splitting out stream into a separate tidal-stream package on the cycseq branch.) |
|
Thanks a lot! |
This is an initial step towards disentangling the controller input logic from the
Streammodule (as discussed in #982). Specifically, it removes the SuperDirt handshake process from the controller thread and instead handles all that per-target using the individual SuperDirt sockets. This adds support for more unusual setups (e.g. one Tidal instance running multiple SuperDirt instances at multiple addresses), but should function identically in most cases.Fixes #793—this uses totally separate addresses for SuperDirt communication and external controllers, so it's not an issue if one is on localhost and the other isn't.
(I've done some tests on my machine, and everything seems to be working the same as in 1.9.4. I started to write some unit tests, but couldn't figure out how to write tests that didn't actually send things over UDP and got nervous that using real UDP communication could result in flaky tests.)