One thing that's missing is error handling for commands that failed to parse. If someone types /gamemode aklwjdlkawlhd, How do I tell them Unknown game mode: aklwjdlkawlhd?

I also noticed that in vanilla, /gamemode aklwjdlkawlhd will give me the "unknown game mode" message while the command is still being typed. What packet is responsible for that, and how do we do that ourselves? It's because the gamemode parser isn't being used in this example.

It is up to whomever handles raw strings to send warnings (the client will show red text with the current parsing). Trying to guess what node a client is talking about with a broken or incomplete command is a bad idea.

But how do I get access to the error that occurred so I can send an error in the chat or whatever else? It looks like the error is thrown away.

https://github.com/valence-rs/valence/pull/446/files#diff-494de62f0e3a6fa84f8d7e1d109af30f30d37abba15f0271dff866d8512c20c1R502-R503

The thing is when trying to match a node an error is assumed to mean that the command doesn't match the given node. For example if we have two executables at /foo and one is a number arg and the other is a string arg /foo bar should not trigger "expected number got bar".

I had a look at the source for GameModeArgumentType and it throws an exception when the gamemode fails to parse.

private static final DynamicCommandExceptionType INVALID_GAME_MODE_EXCEPTION = new DynamicCommandExceptionType(gameMode -> Text.translatable("argument.gamemode.invalid", gameMode));

argument.gamemode.invalid translates to the message the user sees when they type /gamemode aklwjdlkawlhd.

So clearly the error is bubbling up somehow, but I don't know what the semantics are exactly. Might need to dig into the brigadier source.

We would have to make some assumptions 1. no other nodes match 2. no better fit for error reporting. This applies fine to the small set of uncomplicated vinnila commands but anything more complex and dynamic will cause headaches in the future. OTOH We could label exclusive args somehow (the only arg reachable at a certain path) and bubble those errors up.

Maybe vanilla is throwing away all but the last exception? Just a guess, but I'd like to get a confirmation so we can have more confidence in a decision.

How about we hash args based on the path taken to get to them and if the hash is unique then we report errors when the arg does not match. This only solves the problem for unique beaches (most cases). This would not be too hard to implement, I can probably do it in about 30m to an hour tomorrow.

But what is brigadier doing to resolve this situation? I doubt it's anything like that.

I have not read brigadier's code and I am not well versed on java so I don't particularly want to track it down.

Not sure this is the best command to use as an example because people like me will be wondering why we're not just using the normal GameMode parser.

I like how the current version shows multiple paths or redirections. we could add a comment if we must.