[mini.ai][nightly] Conflicts with new LSP keybindings

[igorlfs]

Hello,

Recently, support for incremental selection has been added to neovim. They choose to use an / in ("around node" / "inner node") as the default keymaps for this feature. However, this choice conflicts with mini.ai's around_next / inside_next feature.

I actually wasn't aware mini.ai had this feature (though it is super useful, thanks!), I only found out when I tried using the new LSP selection, and it wasn't working as expected.

Perhaps the defaults should be changed? (I wonder if it also makes sense to change the default around_last / inside_last)

[echasnovski]

Thanks for notifying. I've only became aware of this change a couple of days ago.

Perhaps the defaults should be changed?

Yes, on Neovim's side 🤣

The naming choice is indeed unfortunate for compatibility with 'mini.ai'. I don't think changing defaults in 'mini.ai' is a good idea, because the current keys are the perfect choice.

I think I'd prefer to go the 'miin.operators' route and automatically remap new built-in an / in to something like aN / iN. Alternatively, users themselves can adjust these variants in 'mini.ai'. I'll take a look.

[igorlfs]

to something like aN / iN

Can you provide an option to allow the user to choose said override?

[echasnovski]

Probably, not. The approach is the same as in 'mini.operators': do it by default if an / in will be overridden. If users want to use different keys, they can map directly to them before/after setting up 'mini.ai'. It is a mapping to vim.lsp.buf.selection_range() with a proper argument involving vim.v.count1. It is shown in :xmap an and :xmap in.

[echasnovski]

The conflict is now documented along with the suggested solutions.

I decided to not automatically remap as in 'mini.operators' for several reasons:

• There is no good keys to map to. The aN / iN will conflict with N 'mini.ai' textobjects (which, for example, is recommended for "number" textobject from 'mini.extra'). The grs / grS align with built-in LSP mappings, but conflict with 'mini.operators' "replace" operator default gr keys.
• It is fairly straightforward to manually map LSP incremental selection. So decided to just document how to map it.

Thanks again for bringing attention to this!