Problem:
In LSP configs, the function form of `cmd()` cannot easily get the
resolved root dir (workspace). One of the main use-cases of a dynamic
`cmd()` is to be able to start a new server whose binary may be located
*in the workspace* ([example](https://github.com/neovim/nvim-lspconfig/pull/3912)).
Compare `reuse_client()`, which also receives the resolved config.
Solution:
Pass the resolved config to `cmd()`.
(cherry picked from commit 32f30c4874)
Co-authored-by: Julian Visser <12615757+justmejulian@users.noreply.github.com>
Problem:
Stopping a language server and then calling vim.lsp.start() with the same name/root will return the old language server that's in the middle of shutting down. vim.lsp.start() won't return a new server until the old process has terminated.
Solution:
Introducing a client._is_stopping field that tracks the shutdown phase, preventing the client from being reused.
(cherry picked from commit 0862c1036a)
Problem:
No way to check if a LSP config is enabled without causing it to
resolve. E.g. `vim.lsp.config['…'] ~= nil` will resolve the config,
which could be an unwanted and somewhat expensive side-effect.
Solution:
Introduce `vim.lsp.is_enabled()`.
(cherry picked from commit 03d378fda6)
Problem:
enable() could be more flexible, so that it works even if called "late".
Solution:
- enable(true) calls `doautoall nvim.lsp.enable FileType`.
- enable(false) calls `client:stop()` on matching clients.
This will be useful for e.g. :LspStop/:LspStart also.
(cherry picked from commit 4bc7bac884)
Problem:
Tests call `clear()` even though `clear_notrace()` is already called in
a `before_each()` handler, wasting precious milliseconds!
Solution:
Remove redundant `clear()` calls.
Problem:
In cases when the (in-process) LSP server responds to the request
immediately and calls `notify_reply_callback` the request will still be
marked as pending, because the code assumes that the response will occur
asynchronously. Then the request will be pending forever, because it was
already set as "completed" before we even set it as "pending".
A workaround is to wrap `notify_replay_callback` in `vim.shedule` ([like
so](https://github.com/neovim/neovim/pull/24338#issuecomment-2809568617)]
but that seems counterintuitive.
Solution:
Handle this case in Client:request().
(cherry picked from commit 8315697449)
Problem:
Some language servers do not work properly without a workspace folder.
Solution:
Add `workspace_required`, which skips starting the lsp client if no
workspace folder is found.
Co-authored-by: Michael Strobel <71396679+Kibadda@users.noreply.github.com>
Problem:
`vim.lsp.buf.[implementation|definition|...]({ reuse_win = true })` does not
jump cursor to existing window if buffer is already open.
Steps to reproduce:
1. `nvim repro.lua`
2. Insert anything that lsp can read to open the library definition/implementation, e.g., `vim.keymap.set`
3. open `repro.lua` buffer and the library buffer side by side.
4. type `gd` over `set` to jump to the library definition.
The open buffer is scrolled to the target line, but cursor does not jump.
Solution:
Call nvim_set_current_win if necessary.
(cherry picked from commit 6926fc1615)
Problem:
If a config name contains "*" it causes rtp discovery of `lsp/` to
consider the `*` as a wildcard and could lead to strange and unintended
behaviour. For example, accessing the `'*'` config from a `lsp/` file
would cause an infinite loop.
Solution:
- Explicitly disallow a config name from containing wildcards, with the
exception of `'*'`.
- When Resolving `'*'` config, skip the rtp step.
(cherry picked from commit 2ee896201c)
Problem:
The root dir function is not passed any context and can only assume the
current buffer is the one being attached.
The main use case is for getting the path of the buffer using
`nvim_buf_get_name`.
Solution:
Pass the buffer number as the first argument.
When `root_dir` is a function it can (and often will) call the provided
callback function in a fast API context (e.g. in the `on_exit` handler
of `vim.system`). When the callback function is executed we should
ensure that it runs vim.lsp.start on the main event loop.
Problem:
The floating window for hover and signature help always cuts off a few lines,
because the `_make_floating_popup_size` function counts empty lines as having
zero height.
Solution:
Ensure the height is at least 1.
Allows to retrieve the configuration as it will be used by `lsp.enable`
- including the parts merged from `*` and rtp.
This is useful for explicit startup control
(`vim.lsp.start(vim.lsp.config[name])`)
Closes https://github.com/neovim/neovim/issues/31640
If root_dir is a function it is evaluated when the client is created to
determine the root directory.
This enables dynamically determining the root directory based on e.g.
project or directory structure (example: finding a parent Cargo.toml
file that contains "[workspace]" in a Rust project).
Problem:
An LSP configuration that creates client with no root_dir or
workspace_folders can result in vim.lsp.enable attaching to it multiple
times.
Solution:
When checking existing clients, reuse a client if it wasn't initially
configured have any workspace_folders. This more closely matches the
behaviour we had prior to d9235ef
Design goals/requirements:
- Default configuration of a server can be distributed across multiple sources.
- And via RTP discovery.
- Default configuration can be specified for all servers.
- Configuration _can_ be project specific.
Solution:
- Two new API's:
- `vim.lsp.config(name, cfg)`:
- Used to define default configurations for servers of name.
- Can be used like a table or called as a function.
- Use `vim.lsp.confg('*', cfg)` to specify default config for all
servers.
- `vim.lsp.enable(name)`
- Used to enable servers of name. Uses configuration defined
via `vim.lsp.config()`.
Problem:
str_byteindex_enc could return an error if the index was longer than the
lline length. This was handled in each of the calls to it individually
Solution:
* Fix the call at the source level so that if the index is higher than
the line length, line length is returned as per LSP specification
* Remove pcalls on str_byteindex_enc calls. No longer needed now that
str_byteindex_enc has a bounds check.
Updated the `rpc.connect` function to support connecting to LSP servers
using hostnames, not just IP addresses. This change includes updates to
the documentation and additional test cases to verify the new
functionality.
- Modified `connect` function to resolve hostnames.
- Updated documentation to reflect the change.
- Added test case for connecting using hostname.
Added a TCP echo server utility function to the LSP test suite. This
server echoes the first message it receives and is used in tests to
verify LSP server connections via both IP address and hostname.
Refactored existing tests to use the new utility function.
Problem:
Tests have lots of exec_lua calls which input blocks of code
provided as unformatted strings.
Solution:
Teach exec_lua how to handle functions.
Reverts https://github.com/neovim/neovim/pull/29212 and adds a few
additional test cases
From the spec
> All text edits ranges refer to positions in the document they are
> computed on. They therefore move a document from state S1 to S2 without
> describing any intermediate state. Text edits ranges must never overlap,
> that means no part of the original document must be manipulated by more
> than one edit. However, it is possible that multiple edits have the same
> start position: multiple inserts, or any number of inserts followed by a
> single remove or replace edit. If multiple inserts have the same
> position, the order in the array defines the order in which the inserted
> strings appear in the resulting text.
The previous fix seems wrong. The important part:
> If multiple inserts have the same position, the order in the array
> defines the order in which the inserted strings appear in the
> resulting text.
Emphasis on _appear in the resulting text_
Which means that in:
local edits1 = {
make_edit(0, 3, 0, 3, { 'World' }),
make_edit(0, 3, 0, 3, { 'Hello' }),
}
`World` must appear before `Hello` in the final text. That means the old
logic was correct, and the fix was wrong.
This reduces the number of nil checks around buf_versions usage
Test changes were lifted from 5c33815
Co-authored-by: Mathias Fussenegger <f.mathias@zignar.net>
