As per #14236, performing extmark cleanup in a certain namespace does not guarantee removing all the extmarks inside given namespace. The issue resides within the tree node removal method and results in a couple of rare edge cases. To demonstrate what causes this bug, I'll give an example covering one of the edge cases. === AN EXAMPLE === (A) (B) (C) (D) (E) --------- --------- --------- --------- --------- <0, 1> <0, 1> <0, 1> <0, 1> <0, 1> <0, 2> <0, 2> <0, 2> <0, 2> <0, 2> <0, 3> <0, 3> <0, 3> <0, 3> <0, 3> <0, 4> <0, 4> <0, 4> <0, 4> <0, 4> <0, 5> <0, 5> <0, 5> <0, 5> <0, 5> <0, 6> <0, 6> <0, 6> <0, 6> <0, 6> <0, 7> <0, 7> <0, 7> <0, 7> <0, 7> <0, 8> <0, 8> <0, 8> <0, 8> <0, 8> <0, 9> <0, 9> * * <0, 9> * <0, 9> [0, 10] * [0, 10] <0, 9> [0, 11] [0, 11] [0, 11] [0, 11] [0, 11] [0, 12] [0, 12] * [0, 12] [0, 12] [0, 12] [0, 13] [0, 13] [0, 13] [0, 13] [0, 13] [0, 14] [0, 14] [0, 14] [0, 14] [0, 14] [0, 15] [0, 15] [0, 15] [0, 15] [0, 15] [0, 16] [0, 16] [0, 16] [0, 16] [0, 16] [0, 17] [0, 17] [0, 17] [0, 17] [0, 17] [0, 18] [0, 18] [0, 18] [0, 18] [0, 18] [0, 19] [0, 19] [0, 19] [0, 19] [0, 19] [0, 20] [0, 20] [0, 20] [0, 20] [0, 20] DIAGRAM EXPLANATION * Every column is a state of the marktree at a certain stage. * To make it simple, I don't draw the whole tree. What you see are 2 leftmost parent nodes ([0, 10], [0, 20]) and their children placed in order `MarkTreeIter` would iterate through. From top to bottom. * Numbers on this diagram represent extmark coordinates. Relative positioning and actual mark IDs used by the marktree are avoided for simplicity. * 2 types of brackets around coordinates represent 2 different extmark namespaces (`ns_id`s). * '*' shows iterator position. ACTUAL EXPLANATION Let's assume, we have two sets of extmarks from 2 different plugins: * Plugin1: <0, 1-9> * Plugin2: [0, 10-20] 1. Plugin2 calls `vim.api.nvim_buf_clear_namespace(buf_handle, ns_id, 0, -1)` to clear all its extmarks which results in `extmark_clear` call. 2. The iteration process goes on ignoring extmarks with irrelevant `ns_id` from Plugin1, until it reaches [0, 10], entering state (A). 3. At the end of cleaning up process, `marktree_del_itr` gets called. This function is supposed to remove given node and, if necessary, restructure the tree. Also, move the iterator to the next node. The bug occurs in this function. 4. The iterator goes backwards to the node's last child, to put it in the place of its deleted parent later. (B) 5. The parent node is deleted and replaced with its child node. (C) 6. Since now this node has 8 children, which is less than `MT_BRANCH_FACTOR - 1`, it get's merged with the next node. (D) 7. Finally, since at (B) the iterator went backward, it goes forward twice, skipping [0, 11] node, causing this extmark to persist, causing the bug. (E) ANALYSIS AND SOLUTION The algorithm works perfectly when the parent node gets replaced by its child, but no merging occurs. I.e. the exact same diagram, but without the (D) stage. If not for (D), it would iterate to <0, 9> and then to [0, 11]. So, iterating twice makes sense. The actual problem is in (C) stage, because the iterator index isn't adjusted and still pointing to no longer existent node. So my solution is to adjust iterator index after removing the child node. More info: https://github.com/neovim/neovim/pull/14719
Documentation | Chat | Twitter
Neovim is a project that seeks to aggressively refactor Vim in order to:
- Simplify maintenance and encourage contributions
- Split the work between multiple developers
- Enable advanced UIs without modifications to the core
- Maximize extensibility
See the Introduction wiki page and Roadmap for more information.
Features
- Modern GUIs
- API access from any language including C/C++, C#, Clojure, D, Elixir, Go, Haskell, Java, JavaScript/Node.js, Julia, Lisp, Lua, Perl, Python, Racket, Ruby, Rust
- Embedded, scriptable terminal emulator
- Asynchronous job control
- Shared data (shada) among multiple editor instances
- XDG base directories support
- Compatible with most Vim plugins, including Ruby and Python plugins
See :help nvim-features for the full list!
Install from package
Pre-built packages for Windows, macOS, and Linux are found on the Releases page.
Managed packages are in Homebrew, Debian, Ubuntu, Fedora, Arch Linux, Gentoo, and more!
Install from source
See the Building Neovim wiki page for details.
The build is CMake-based, but a Makefile is provided as a convenience. After installing the dependencies, run the following command.
make CMAKE_BUILD_TYPE=RelWithDebInfo
sudo make install
To install to a non-default location:
make CMAKE_INSTALL_PREFIX=/full/path/
make install
To inspect the build, these CMake features are useful:
cmake --build build --target helplists all build targets.build/CMakeCache.txt(orcmake -LAH build/) contains the resolved values of all CMake variables.build/compile_commands.jsonshows the full compiler invocations for each translation unit.
Transitioning from Vim
See :help nvim-from-vim for instructions.
Project layout
├─ ci/ build automation
├─ cmake/ build scripts
├─ runtime/ user plugins/docs
├─ src/nvim/ application source code (see src/nvim/README.md)
│ ├─ api/ API subsystem
│ ├─ eval/ VimL subsystem
│ ├─ event/ event-loop subsystem
│ ├─ generators/ code generation (pre-compilation)
│ ├─ lib/ generic data structures
│ ├─ lua/ Lua subsystem
│ ├─ msgpack_rpc/ RPC subsystem
│ ├─ os/ low-level platform code
│ └─ tui/ built-in UI
├─ third-party/ CMake subproject to build dependencies
└─ test/ tests (see test/README.md)
License
Neovim contributions since b17d96 are licensed under the
Apache 2.0 license, except for contributions copied from Vim (identified by the
vim-patch token). See LICENSE for details.
Vim is Charityware. You can use and copy it as much as you like, but you are
encouraged to make a donation for needy children in Uganda. Please see the
kcc section of the vim docs or visit the ICCF web site, available at these URLs:
http://iccf-holland.org/
http://www.vim.org/iccf/
http://www.iccf.nl/
You can also sponsor the development of Vim. Vim sponsors can vote for
features. The money goes to Uganda anyway.
