<80>kb character in notifications

[pBorak]

Describe the bug

Hey, I've noticed that this changes introduced weird behavior when interacting with vim-fugitive.
After doing a git commit within the nvim with the use of vim-fugitive there's an extra <80>kb character present in the notification. See screenshots below

To Reproduce
Steps to reproduce the behavior:

1. Open fugitive popup
2. Stage files
3. hit cc to open the commit window
4. write commit-msg and hit wq

Expected Behavior
There's no <80kb>

Screenshots

Before 8463b0 it looked this way

[Pheon-Dev]

@folke , can't seem to figure out why issue #142, here, and this one kind of have some sort of problem something in common. I've been facing the same problem too whereby the [SHIFT] key seems to be already pressed then a notification appears whenever the key is actually pressed showing what would have been the result of [SHIFT + ,] which is [<] as illustrated below.
Note: No one pressed the [SHIFT] key, just the [,] key which in itself I mapped it to saving a file(wasn't really using it for [?] search next :-)

[pBorak]

👍🏻 I have also noticed the issue with < notifications

[folke]

Check :h shortmess. t option. Standard Neovim behavior

[folke]

@Pheon-Dev #142 is an nvin-notify problem, not Noice.
Your issue #99, is not an issue at all. Your issue #99 was because of a plugin.
All unrelated to this.

This is likely caused by my hack to make keys work in cmdline during substitute. But haven't yet found anything that could cause that.

[folke]

Just looked it up and <80>kb is indeed the internal code for <backspace>.
Which I do here

noice.nvim/lua/noice/util/hacks.lua

Line 255 in ac29174

vim.api.nvim_input("<space><bs>")

[Pheon-Dev]

Thanks for the explanation, it makes more sense now. I'll try to config my nvim-notify a bit better to avoid receiving unnecessary messages

[folke]

Once the upstream issue below is solved, this issue can be solved

• redraw doesn't update the UI during substitute neovim/neovim#20463

[folke]

I was able to work around the cursor issue by calling update_screen through ffi, so your issue should no longer happen