Dani Dickstein

This affects the default LSP handler for `window/showMessage`, which uses `nvim_out_write`. Should that be changed to `nvim_echo` while this issue remains open?

The displaying of semantic tokens even when a client wishes to opt out by indicating that it is not capable of displaying them.

In general, it's misbehavior to send messages without checking that the client has the appropriate capability (and likewise, for the client to send the server messages without checking server capabilities).

I've updated the original post to clarify. It is sad that we can't add an annotation in `[%sexp x]` - after all, adding the annotation will make it valid!

(The latter two obviate the [recently added `advance` function](https://github.com/ocaml/ocaml-lsp/pull/1086))

A couple questions on the above discussion: - @oblitum drew a distinction between _concealing_ and _inlaying_. I'm not sure I understand the significance of the distinction or what this means...

This happens because `SuperTabCodeComplete` has the line: ```viml let Func = function(b:SuperTabChain[0]) ``` which makes a `funcref` out of `b:SuperTabChain[0]`, and the [docs for `v:lua`](https://neovim.io/doc/user/lua.html#v%3Alua-call) explicitly say: > Note: v:lua...

What's the right mental model for the current semantics of `nvim_win_set_buf` re: side-effects that do and don't happen? And with the changes planned to address this issue and https://github.com/neovim/neovim/issues/13201, what...

Not able to build it yet on the machine where this crash happened. How long after a minor release (in semver terms) will the previous minor release be unsupported?

Didn't test with appimage, but did test on my personal machine which runs 0.8.2 and was able to reproduce the problem as described, so I don't think it's been fixed...