method setEditorInfo is not supported by any of the servers registered for the current buffer

[gpncarl]

https://github.com/zbirenbaum/copilot.lua/blob/b16f659a55f68a53b407dbc34e52456c4ec41f26/lua/copilot/copilot_handler.lua#L53

why not send_editor_info on_attach?

[kozer]

I get this as well

[zbirenbaum]

@gpncarl It was in on-attach, but moving it to the end of on-init means that it sends once to the rpc (as it should) after the initial attach. If it is in on-attach, every time a new buffer is opened and copilot attaches to it, the request is resent.

I believe I fixed this as a response to another issue yesterday. Please packer sync and let me know if you can reproduce it.

[gpncarl]

@zbirenbaum I still call reproduce it

vim.lsp.buf_request_sync in function send_editor_infoneed buffer attached first but on_attach is called later than on_init

[zbirenbaum]

Yes, on_init is called before on_attach, but we don't need to send it every on_attach. Take a look at the code for on_init:

    on_init = function(client, _)
      vim.lsp.buf_attach_client(0, client.id)
      if vim.fn.has("nvim-0.7") then
        vim.api.nvim_create_autocmd({ "BufEnter" }, {
          callback = function()
            util.attach_copilot()
          end,
          once = false,
        })
      else
        vim.cmd("au BufEnter * lua require('copilot.util').attach_copilot()")
      end
      send_editor_info()
    end,

vim.lsp.buf_attach_client(0, client.id) is the very first thing called. At this point, on_attach will be called. on_attach will also be called every time the autocmd subsequently triggers. send_editor_info() is called only once, at the end of on init, which is well after vim.lsp.buf_attach_client(0, client.id) is called. I can't reproduce this behavior at all, and I am very confused as to why it is doing it for you.

If you have a moment, I would like to try doing a vim.schedule on it and see if that fixes your issue.

[zbirenbaum]

Ya know what, it seems to function without sending editor info, and considering it is hardcoded, I'm just going to remove the call to it for now. Scheduling should fix the issue but it's not worth the effort for something that is probably only there for Micro$oft's telemetry....

[zbirenbaum]

It's gone now. Hopefully this doesn't cause any other issues but I doubt it will honestly.