Add a method to find builtins with server name.

[DoeringChristian]

Issues

• I have checked existing issues and there are no existing ones with the same request.

Feature description

Hi,

I am using Mason to install null-ls servers with the plugin mason-null-ls.nvim.
This plugin gives me a string with the name of the installed server. I would like to automatically enable all builtins for that server. When testing if there exists a formatter, code_action etc using:

if null_ls.builtins.formatting[source_name] ~= nil then
    null_ls.register(null_ls.builtins.formatting[source_name])
end

I get a warning if the server does not support formatting.
Is there a way to test if this builtin does exist?
If not could this be implemented?

Thanks for your Help.

Help

Yes, but I don't know how to start. I would need guidance

Implementation help

No response

[jose-elias-alvarez]

Sorry, I'm not sure I understand the request. Don't the source names from mason-null-ls match the source names here?

[DoeringChristian]

They do but when indexing the builtin tables to lookup weather that source does exist and the source does not it generates a warning here.

[jose-elias-alvarez]

Right, so if the names match, then what is the issue? At what point do you access a built-in that doesn't exist?

The existing warning is important to prevent issues caused by typos, and I'd be hesitant to add a method for such a specific use case.

[DoeringChristian]

The problem is that I don't know if the source supports formatting, code_actions etc. I would need to test if the source is in builtin.formatting, builtin.code_action etc.
example.

[jose-elias-alvarez]

I see. This is something we could add here, but I think a better solution might be to have mason-null-ls pass more information than just the source's name to user-defined handlers. As you've seen, the name alone is not enough context to properly handle the source, especially since sources like ESLint have the same name but 3 different methods.

[DoeringChristian]

That makes sense I think having both would be useful nonetheless. If for example the source supports something that has not yet been implemented in null-ls' bultins one could check if the "feature" exists there.

[jose-elias-alvarez]

I think there's some confusion about terminology here because I'm not sure what you mean by "source" (maybe Mason has its own definition?). If I understand correctly, you want to know if a given command (e.g. eslint) is supported by null-ls, and if so, you want to know which methods it supports. Is that right?

If so, it's a pretty niche use case, but it's not something that's easy to do right now, so I think we can support it.

Edit: I originally wrote an example of an approach we could use, but the issue with that approach is that the name of null-ls source doesn't always match the name of the command. Honestly I think your best solution is to copy this logic instead of indexing the builtins metatable to avoid the error.