Add simple option to control rust-analyzer insert vs replace completion behavior

[ZXY595]

Checked other resources

• This is not a bug report or a question (Please use issues for those).
• I have searched existing discussions.
• I have searched existing issues.
• This cannot be implemented in rust-analyzer.
• I have added a very descriptive title to this feature request.

Feature description

Hi 👋,

I’d like to ask whether it would make sense to add a simple configuration option in rustaceanvim to control whether rust-analyzer prefers insert or replace behavior for completion items.

Background

On Neovim 0.11.5, rust-analyzer completion defaults to replace instead of insert.

After digging through Neovim and rustaceanvim source code (nvim capabilities, nvim completion, rustaceanvim server capabilities, rustaceanvim config), issues (rust analyzer issue, rust-analyzer discussion), and PR , I finally found a way to switch this simple behavior is via a relatively verbose LSP capability override, like this:

vim.g.rustaceanvim = {
  server = {
    capabilities = {
      textDocument = {
        completion = {
          completionItem = {
            insertReplaceSupport = true,
          },
        },
      },
    },
  },
}

In short: manually injecting insertReplaceSupport into the server capabilities just to control this one behavior.

Neovim upstream status

This workaround is temporary by nature. In Neovim’s latest development version:
After Neovim enable insertReplaceSupport, make_client_capabilities() already includes
insertReplaceSupport = true by default.
So on future Neovim releases, this configuration becomes unnecessary.

Question / proposal

Would it make sense to:

Add a simple rustaceanvim option (e.g. something like server.insert_replace = true/false) that internally adjusts insertReplaceSupport, and later deprecate/remove it once Neovim stable includes the upstream fix?

or

Do nothing on rustaceanvim’s side and simply wait for Neovim to release the fix?

Personally, I feel a small, explicit option could greatly improve UX for users on current stable Neovim, without requiring them to understand LSP capability internals, and this has been annoying me for months. — but I’m very interested in the maintainer’s perspective on whether this fits the project’s scope.

Thanks a lot for maintaining rustaceanvim 🙏

[mrcjkb]

Hey 👋

Completion isn't rustaceanvim's responsibility.
For that reason, I recommend to wait for the neovim fix (or rust-analyzer fix, since the issue is still open - whichever arrives first).

In the meantime, if you don't want to wait, you can use your workaround or propose adding the insertReplaceSupport capability via your completion plugin.