Christian Clason

Sorry, but for "it breaks plugins" to hold true, it suffices to break _some_ plugins. And for various reasons, `tpope` plugins are of prime concern here -- both as the...

Patching fails on Windows, otherwise looking good \o/ I think we're OK with bumping the minimal required version of libvterm to 0.2 now, so we don't need the `#ifdef` guard...

@cryptomilk I couldn't wait so I pushed the changes to your branch -- your patch and the libvterm-0.3-RC1 release passed all tests (I had to drop the VLA patch for...

> But if you want to commit this for now on the expectation that you'll bump it soon then that's fine. Yes, exactly. This may help surface bugs that can...

(@carlocab also means that you can finally get rid of that resource in the Neovim formula -- to make up for the treesitter parsers ;))

> Does this work with libvterm 0.2, though? I'm already predicting multiple issues opened about a broken brew install --HEAD neovim. Any more than we're already getting because people have...

Should be documented in * https://github.com/neovim/neovim/blob/master/CONTRIBUTING.md#clang-sanitizers-asan-and-ubsan * https://github.com/neovim/neovim/tree/master/src/nvim#build-with-asan mentioning the CMake version requirement.

Alternative, less intrusive approach: 1. Add a note to the `uri_from_fname` docstring about relative paths not being validated and possibly leading to invalid URIs. 2. Change `:h` to `:p:h` in...

That sounds like a bug of `fnamemodify`... But OK, not going to debate this.

Yeah, it's even documented. Weird.