user defined is_available

[lukas-reineke]

sources can define the is_available function, but there is no way for users to do this.

I think it would be nice to expose this to users as well. So sources don't have to build custom solutions for this.

require("cmp").setup({
	sources = {
		{
			name = "foo",
			is_available = function()
				-- user logic
				return false
			end,
		},
	},
})

I think it should not overwrite the source defined is_available. cmp should check both. If either one returns false, the source is not available.

I can make a PR if you want this

[hrsh7th]

Currently, I think I have to clean up the source configuration API. (But it's difficult.)
I would appreciate any ideas.

[dmitmel]

I think it should not overwrite the source defined is_available. cmp should check both. If either one returns false, the source is not available.

What about passing plugin's original is_available function to user's is_available, like this:

cmp.setup({
  sources = {
    {
      name = 'rg',
      is_available = function(original_is_available_fn, ...)
        local is_in_git_tree = vim.trim(vim.fn.system('git rev-parse --is-inside-work-tree')) == 'true'
        return is_in_git_tree and original_is_available_fn(...)
      end,
    }
  }
})

? This way the user gets to decide whether the source-defined is_available rule should be used or not.

[hrsh7th]

sources = {
  {
    name = 'buffer',
    override = {
      get_keyword_pattern = function(source, ...)
        return source:get_keyword_pattern(...)
      end,
      get_trigger_characters = function(source, ...)
        return source:get_trigger_characters(...)
      end,
      complete = function(source, params, callback)
        source:complete(params, function(response)
          -- modify responses...
          callback(response)
        end)
      end
    }
}

[hrsh7th]

I think if we introduce this, we should deprecate the keyword_pattern option etc.

[lukas-reineke]

overwrite system makes sense to me 👍

It might be easier to maintain if it just works as a sort of constructor function? Instead of override being an object, it is a function that gets source as the argument.
And it returns a new source. (then the new source is set as __index)

This way, users can do whatever they want to the source object. And you never have to touch this part again, even if you add more things to sources.

sources = {
    {
        name = "buffer",
        init = function(source)
            -- I can do more stuff here.
            return {
                get_keyword_pattern = function(...)
                    return source:get_keyword_pattern(...)
                end,
                get_trigger_characters = function(...)
                    return source:get_trigger_characters(...)
                end,
                complete = function(params, callback)
                    source:complete(params, function(response)
                        -- modify responses...
                        callback(response)
                    end)
                end,
            }
        end,
    },
}