Problem: values from completionList.itemDefaults always replace the
item's own commitCharacters and data. Since 3.18 a server can ask for
them to be merged instead.
Solution: Handle CompletionList.applyKind and advertise applyKindSupport.
Drop itemDefaults once folded into the items: get_items() runs twice on
the same result and a Merge isn't idempotent.
https://microsoft.github.io/language-server-protocol/specifications/lsp/3.18/specification/#completionItemApplyKinds
Problem:
`nvim__ui_detach` was added in 85e0559d46 in order to implement
`detach_others` from Lua. Using Lua in this case is doing more harm than
good.
Solution:
Extract `ui_detach_channel`, which also allows it to be used for the
"self detach" path from `ex_detach`.
Bonus: the old MSWIN path always called `os_swap_to_hidden_console()`;
now `ui_detach_channel` only does so for a stdio channel. For the common
stdio TUI this is identical; for a socket UI it's more correct (no
parent console to swap).
Problem:
Third-party directory browsers do not have a documented way to take over local
directory buffers from nvim.dir.
Solution:
Document deleting the nvim.dir augroup and handling FileType directory as the
after-startup handoff, and avoid installing the built-in buffer-local "-" mapping
when "-" is already mapped.
Problem: LSP completion commitCharacters are not handled. Typing a
commit character (e.g. `.`, `(`, `;`) while a completion item is
selected does not accept the item first.
Solution: Store commit characters as a flat string in complete-items.
Check it before completion stops, accept the match and let
the character be inserted normally.
Problem: Every LSP capability that sends a request on a document change
to update its state was using its own buffer-local autocmd to do so.
That means there is a separate autocmd per buffer per active
capability/feature, and the actual code inside the autocmd callback was
virtually identical. The same problem occurred for capabilities that had
components to draw on the screen that had their own decoration providers.
Solution: Introduce a central `LspNotify` autocmd and decoration
provider in the capability module itself. New base class methods
`on_close` and `on_change` that take a client_id have been provided, for
`didClose` and `didOpen`/`didChange` notifications respectively. New
base class method `on_win` that takes topline and botline has been
provided to add extmarks or perform other work when lines of a buffer
are being drawn.
The autocmd callback loops through each active capability instance for
the buffer and calls a corresponding method with the triggering
client_id if it is currently attached to the buffer that triggered the
autocmd. The decoration provider just calls on_win for all active
capabilities on the buffer.
This slightly tweaks folding range to make use of these new client-id
specific callbacks.
Problem
The current list of properties for inlay hint resolve support includes a
non-existent "location". This field should be "label.location" according
to the specification. Tooltips and commands for inlay hint "parts"
aren't advertised as being resolvable.
Solution
Remove "location" and add "label.location", "label.tooltip", and
"label.command".
Closes#40740
Problem:
- No UI for listing extmarks.
- `:marks` doesn't support cmdline-completion.
Solution:
- Add completion and extmarks support to `:marks`.
- Update `get_buf_local_marks` to include the ":" (prompt-buffer) mark.
Problem:
This functions are most likely never used in the wild.
- They are redundant with `nvim_get_context` + `nvim_load_context`.
- The "context" concept was never fully developed and hasn't gained
traction.
Solution:
Drop these vimscript aliases, they are just extra cruft.
Problem:
After 25439dad7e, `opts` is optional, but not params that follow it.
Solution:
- Update `gen_api_dispatch.lua`, docs, tests.
- Enhance the `nvim_get_api_info()` api-metadata to include a boolean
flag indicating whether a parameter is optional:
```
[type, name, optional]
```
Note: Currently, optional params are assumed to be Dict/Array. Since we
don't yet have any use-cases of optional params following `opts`, the
build just checks this assumption and we can deal with it later if we
ever care to:
nvim_get_hl: optional param "foo" has type "Integer" but (currently)
we assume Dict/Array
Problem:
Continue separators normalization, and try to keep it at the
nvim <-> external boundary, e.g., `fn.xxx`, `api.xxx`, `:xxx`
Solution:
some key changes
- normalize `$HOME-windows`
- normalize entry points of changing directory
- `nvim_set_current_dir()`, `chdir()`
- normalize the named pipe
- `--listen` arg, `--server` arg, `:restart`
- `serverstart()`, `sockconnect()`, `serverstop()`
- make `expand()` respect 'shellslash' again
- clean up `did_set_shellslash`
- replace `forward_slash` with `TO_SLASH`
- remove obsolete `TMP_PATHSEPSTR`
- remove `slash_adjust` in `do_autocmd_dirchanged`?
- make `fnamemodify()` always return `/` (except when substituting
separators via `:s`, `:gs` ?)
Note:
- these funcs still apply `slash_adjust` on return, as before:
`:pwd`, `chdir()`, `exepath()`, `getcwd()`
- these funcs alwarys return `/`, unlike before the normalization PRs
- `getcompletion()`, `finddir()`, `findfile()`
Also
- clean up code, comments, formatting and tests
- move `TO_SLASH` from `f_chdir` to `changedir_func`
- move `TO_SLASH` from `f_bufadd` to `buflist_new`
Co-authored-by: Justin M. Keyes <justinkz@gmail.com>
Problem:
Malformed server responses with `null` (vim.NIL) values, do not show
clear error messages.
Solution:
Assert two known, specific cases, to avoid user confusion.
Problem:
URI buffer names can end with a slash, but slash-ended directory
matching should not preempt URI filetype patterns.
Solution:
Do not use the slash shortcut for names with URI schemes, so those
continue through normal filetype matching.
Co-authored-by: Adam Karafyllidis <akisarou90@gmail.com>
The purpose of `v:echospace` was to avoid disruptive "Press ENTER"
prompts. Since those are no longer a problem with ui2, the complexity
of maintaining it up-to-date is no longer justified.
Problem: a long message is cropped to keep the ruler visible. But when
the message is repeated, it can be written over the ruler, while the
repetition indicator "(1)" is still placed just before the ruler, which
ends up somewhere inside the message.
Solution: currently, long messages are only cropped when setting the
'last' virttext. Make sure this happens after a repeated long message.
Problem: when the search count is displayed, the showcmd virt_text is
set to 11 empty spaces to ensure a consistent distance between search
count and ruler, or screen edge in case of noruler. But when the search
count is removed to make place for a message, the empty dummy showcmd
remains and unnecessarily erases part of the message.
Solution: properly remove showcmd together with search count.
Problem: ui2's ruler is no longer of fixed width, and the internal width
variable `ru_wid` now acts as an arbitrary maximum, which is unnecessary
because the same can now be achieved transparently with an item group.
Furthermore, the old way of setting `ru_wid` with an overloaded subset
of the item group syntax leads to a number of inconsistencies. In
particular, when the configured ruler expression is reused for including
the ruler into the statusline, where the item group syntax is
interpreted in the usual way. The fact that this width is set in the
`minwid` position is more confusing now that it is a de-facto maximum.
Solution: for ui2, constrain the ruler only by half the screen width,
which previously was an additional limit relevant on small screens.
Skip the special-case treatment of the ruler's leading item group spec,
process it like a normal item group.
Example: the current fallback ruler expression in the default statusline
now actually works for `rulerformat` itself:
set rulerformat=%-14.(%l,%c%V%)\ %P
because the result can naturally expand to 18 cells instead of being
truncated to the default 17 cells, and removing the period after
`minwid` doesn't change the result, whereas before, it would unhelpfully
lead to setting the fixed ruler width to 14 cells.
BREAKING CHANGE: if the user used to set the ruler's fixed width, it
will now be treated as minimum instead of a maximum width, which in
practice increases backwards-compatibility with ui1 and vim. The main
remaining difference is that item groups are right-aligned by default,
but backwards-compatibility can be achieved by setting a negative
`minwid`, which ui1 and vim ignore. To restore the previous fixed width,
specify `maxwid` as well, but this does not work in ui1 and vim.
If `rulerformat` contains a `%=` without enclosing item group, it will
now expand to fill half the screen width. But in that case, it will
already spread out similarly in the new expression-based statusline, so
consistency of the ruler in both positions is increased.
To simulate the previous default behaviour, wrap it in `%-17.17(...%)`.
For best compatibility and reuse with vim, leave out `maxwid` and make
`minwid` negative, e.g. `%-20(...%)`.
fix https://github.com/neovim/neovim/issues/40247
Problem: nvim_cmd() accepts number/boolean in args and converts them
to strings, but the keyset declares ArrayOf(String).
Solution: use ArrayOf(Union(Integer, String, Boolean)) and wrap union
types in parens when generating array annotations.
Problem:
`vim.filetype.match()` needs a cheap way to recognize directory buffers
without doing filesystem stat work.
Solution:
Ensure full buffer names for directories end in a trailing slash. Now
directory buffers can proceed through the normal 'filetype' path.
Note side-effects: session and ShaDa buffer-list restore behavior must
be compatible, so those + corresponding tests must be updated.
Problem: No concise way to execute a callback with temporarily set
working directory. This might be useful when sourcing nested files to
allow them to assume that current working directory is their root
directory.
Solution: Add `cwd` context to `vim._with().`
Problem: when an LS client detaches from the buffer, only pull diagnostics
are cleared via capability framework. Push diagnostics remain stuck even
when client stops/restarts.
Solution: clear push diagnostics on client detach.
ref #33864
When a server supports both document and workspace pull diagnostics,
`on_refresh` only dispatched a `workspace/diagnostic` request. The
workspace response handler skips buffers with `pull_kind == "document"`
(i.e. all buffers opened by the user), so their diagnostics went stale
until the next `didChange` or `didOpen` event.
Change `on_refresh` to always refresh document-pull buffers via
`textDocument/diagnostic`, regardless of whether the server also
supports workspace diagnostics. This ensures that opened buffers
see updated diagnostics (e.g. after a save triggers an external
tool like PHPStan) without requiring the user to re-enter insert
mode.
Problem:
`vim.fs` does not provide a directory creation helper matching its
filesystem API shape.
Solution:
Add `vim.fs.mkdir()` as a thin wrapper around `vim.fn.mkdir()`, with
`parents` and `mode` options.
Problem: The codelens LSP module was using its own raw buffer events and
its own debounce mechanism for refreshing code lens in attached buffers.
Solution: Switch the module to using the LspNotify autocmd events.
LspNotify fires just after document versions are synced with the server
and provides a built in debounce mechanism for changes.
Additionally, this fixes some bugs with the previous implementation:
1. The workspace/codeLens/refresh handler re-requested codelens for all
buffers but when the response came back, it forced an extra redraw
after clearing the work the handler had just done.
2. Document synchronization was reworked to be more resilient to
multiple clients providing codelens for a single buffer. The latest
document version is now separately tracked per client (and per
client's lenses per row) instead of for the buffer as a whole. This
allows the on_win() function to properly redraw all codelens even
when different clients' responses for a particular document version
come back at different times.
Problem: :Man unloads hidden buffers, so reopening large pages reruns rendering
even when the width is unchanged.
Solution: keep regular man buffers loaded and refresh them only when the
effective width changes.
Problem:
Inlay hints used separate global and per-buffer bufstates tables and
bespoke global autocmds for managing the inlay hint state across buffers
and clients, duplicating the lifecycle logic already provided by the
Capability framework. This caused inconsistencies in how client state
was handled and inlay hint state lifecycle was managed compared to other
LSP features.
Solution:
Replace the ad-hoc bufstate tracking and global autocmds in
vim.lsp.inlay_hint with a proper InlayHint subclass of Capability.
This also refactors the way inlay hint state is managed and fixes bugs I
found while doing this:
1. For each line with inlay hints, the list of the hints along with
whether they have been applied is stored in a current result on the
client state. This allows the on_win decorator to clear all inlay
hints for an old document version once, and then re-add the new
version's hints line-by-line as they are drawn to the screen,
modeling the semantic tokens module.
2. It fixes problems with mixing results from multiple clients attached
to the buffer by fully moving each client's state to its own table.
Previously, only the most recent document version used to populate a
line's inlay hints was stored, but there was no distinction for which
client the hints may have come from. (Fixes#36318)
3. It fixes the workspace/inlayHint/refresh server->client notification
behavior. Previously it would only re-request inlay hints for buffers
currently displayed in a window but would not invalidate them in
non-displayed buffers (or provide any mechanism for those buffers to
re-request at a later time). Model semantic token module here again
by invalidating all buffers, and adding a BufWinEnter autocmd to
refresh hints.
4. Add a mechanism to cancel in-flight requests if a new request for a
newer document version is made before the last one returned
5. Handle stale results by simply dropping them.
Ref #6645
Problem:
When a window is resized it takes space from the window right/below first,
and only falls back to the window left/above when there is no more room.
Sometimes a user wants the space to come from a specific direction.
Solution:
Add nvim_win_resize(win, width, height, {anchor}) which resizes a window
with a choosable anchor edge, letting a window grow leftwards or upwards
by taking space from the window to the left or above first. The default
anchor reproduces nvim_win_set_width()/nvim_win_set_height().
Problem:
`vim.lsp.buf.format()` accepts ranges using nvim indexing, where an
end column of -1 means end of line. LSP ranges cannot use that,
which is confusing for things like range formatting.
Solution:
Resolve -1 end columns to the line length before converting the range to
LSP positions.
Problem:
When buffers are by default `nomodifiable`, such as when Nvim starts with
`-M`, the health buffer cannot be updated.
Solution:
Always set `modifiable` before modifying the buffer.
Problem:
Plugins using RPC sockets cannot detect when the peer closes a
`sockconnect()` channel, so reconnect logic has no reliable trigger.
Solution:
Add a `ChanClose` event with channel info before the channel is removed,
matching the existing `ChanOpen`/`ChanInfo` event model.
Problem: reset_timer() was being called without checking for whether the
client state for the client_id still existed. debounce_request() starts
a timer that defers a call to send_request() which then calls
reset_timer(). If the timer fires after the client_state is erased, then
the deferred function attempts to dereference the timer on a nil client
state.
Solution: Change reset_timer to take a state directly so it can't be nil
and move the reset_timer() call inside a guard that ensures state
exists. Additionally, reset a client's timer when the client detaches so
it doesn't become dangling.
Problem: The document_color lsp module was already using the capability
framework but was still using raw buffer events to handle requests and
reloading. This means that every keystroke was sending a document_color
request to the server since there was no debounce in the raw handlers.
Solution: Switch to using LspNotify autocmd events. LspNotify fires just
after new document versions are synced with the server and provides a
built in debounce mechanism for changes. It also provides the signal for
when the current state should be cleared (didClose). The detach part is
already handled by the capability framework.
Fixes#39785
Problem:
`request()` and `notify()` are methods of the object returned by
`vim.lsp.rpc.start()`/`connect()`, but were rendered with module-level
helptags (`vim.lsp.rpc.request()`, `vim.lsp.rpc.notify()`) (erroneously
implying module functions that do not exist).
Solution:
Mark the wrappers `@private` and describe them on `vim.lsp.rpc.Client` instead.
