This was initially added so that the __NVIM_DETACH environment variable
can be added to the jobs started in this test. That environment variable
is no longer needed, and there is also vim.tbl_extend() that can be used
to add an environment variable to a job anyway.
Also, make a shallow copy of opts.env in setup_child_nvim(), as mutating
the opts.env passed in may mask problems in other tests.
(cherry picked from commit f876fd906e)
Problem: A wrapped command line and screen width may be redrawn
repeatedly after calls to `redrawstatus` from lua.
Solution: redrawcmd() redraws the command line, but msg_clr_eos() may
invalidate cmdline_was_last_drawn during the redraw process.
Restore cmdline_was_last_drawn when redrawcmd() completes.
Co-authored-by: Sam Reynoso <sam@codeoutpost.com>
Co-authored-by: Luuk van Baal <luukvbaal@gmail.com>
(cherry picked from commit 6be9459d35)
Problem: No message kind and multiple events for :recover and
(non-prompt) swapfile attention messages.
Solution: Assign these the "list_cmd" and "wmsg" kind.
Problem: The "Scanning:" completion, bufwrite, and indent (there may be
more) messages which indicate progress can use the "progress" kind
for their msg_show event. Indent message does not have a kind.
Solution: Emit these messages with the "progress" kind. Set the message id
to the replaced kind so that a UI knows to replace it (and to provide
a migration path in case a UI was distinguishing these messages for
whatever reason).
Problem: The refactor to use Capability left around some cruft and
semi-broken configuration for debounce.
Solution: Clean up now-unnecessary helper methods and simplify
deprecated ones to pass through to the non-deprecated ones. `debounce`
now defaults to 200 for all buffers but is overridable via the
deprecated start() method, which continues to take the max value
specified for any client attached to the buffer.
If we wish to expose changing the debounce in a non-deprecated way, we
will need to consider a "configuration" function, or even a bespoke
method to set the debounce time on the main metaclass (or provide
options to override for a particular buffer). General configuration of
specific LSP features is an as-of-yet unsolved problem.
(cherry picked from commit 859790e244)
Problem:
A colored (guisp) underline showing through a 'winblend' float
lost its special color and followed the foreground instead.
Solution:
In the blend-through case, blend the underline's special color
only when the cell below sets sp explicitly; otherwise clear it.
(cherry picked from commit 823517ad1d)
feat(lsp): use LspNotify for semantic tokens
Problem: The semantic token module is using its own debounce timer for
the buffer on_lines event. If its internal debounce is shorter than the
changetracking module's debounce, it's possible for semantic token
requests to fire for changed buffers before the textDocument/didChange
notification is sent to the server.
Solution: Trigger semantic token requests from the LspNotify autocmd
when the method is the didChange or didOpen notifications, which
enforces a strict happens-before relationship for the sync change
notification followed by a semantic token request.
Note: There is still an internal debounce mechanism in the semantic
token module to handle other debouncing needs specific to its
functionality, such as debouncing server refresh notifications and
handling WinScrolled events when using range requests.
Co-authored-by: jdrouhard <john@drouhard.dev>
Problem:
A text edit positioned entirely past the last buffer line, with
newText ending in a newline, leaves a stray blank line: the
past-the-end path appends the trailing empty fragment produced by
vim.split() and does not set has_eol_text_edit, so the end-of-buffer
cleanup is skipped. Formatting servers emit such edits whenever
formatting moves text to the end of a document.
Regression from ec94014cd1 (#20137), which split the past-the-end
fast path off the clamp path that sets the flag.
Solution:
Set has_eol_text_edit in the past-the-end path, like the adjacent
path that clamps end_row.
(cherry picked from commit d42f7ee9dc)
Co-authored-by: Aaron Tinio <aptinio@gmail.com>
Problem:
When a server sends workspace/codeLens/refresh while an automatic codelens
request is already scheduled, Nvim ignores the server refresh. This can leave
rendered codelens text stale until another buffer edit triggers a new request.
Solution:
Cancel the pending automatic request and send the server-requested refresh
immediately. This preserves request coalescing while giving explicit server
refreshes priority.
Co-authored-by: Tristan Knight <admin@snappeh.com>
Problem:
Various out-of-bounds writes inherited from vim (examples assume MAXPATHL==4096):
- truncated item groups where minwid>maxwid:
nvim --clean +"set stl=%<%{%repeat('x',4096-11)%}%50.5(12🙂345%)"
leads to fillchars spilling over the end of the group/buffer while trying to
compensate for truncating at a multicell character because minwid<=maxwid is assumed
- left-aligned item groups with multi-byte fillchar:
nvim --clean +"set fillchars+=stl:∙ stl=%<%{%repeat('x',4096-3)%}%-2(X%)"
wrongly leads to padding at the end of the statusline and `out_p-out==4097`
because the bounds check assumes a 1-byte fillchar
- right-aligned item groups with 1-byte fillchar:
nvim --clean +"set stl=%<%{%repeat('x',4096-4)%}%4(XY%)"
leads to "YX" instead of "XY" at the end of the statusline
because `memmove` is done before adjusting the offset
- right-aligned item groups with multi-byte fillchar:
nvim --clean +"set fillchars+=stl:∙ stl=%5(X%)"
leads to "∙∙∙∙<e2>", i.e. the fillchar is being written over the group contents
and eventually being overwritten itself at the second byte with the final NUL,
because the padding counter assumes a 1-byte fillchar; to crash Neovim,
nvim --clean +"set fillchars+=stl:∙ stl=%<%{%repeat('x',4096-149)%}%50(X%)"
Solution:
Clearer variable names and no recycling of variables for different purposes.
(cherry picked from commit e36659c82f)
Problem:
`foldcolumn` is empty for virtual lines above the start of a nested
fold.
Solution:
For virtual lines, compute the outer fold level and display it by
reusing the logic from `fill_foldcolumn`.
(cherry picked from commit fe154f4d45)
Problem:
stdpath() may return a DOS 8.3 "shortened" filename, because Windows
truncates some long usernames into `6ch~N` names. Then features such
as `nvim_get_runtime_file` fail to find the file.
Analysis:
When expanding an 8.3 filename path like `C:/Users/ADMINI~1/AppData/*`,
we treat `~` as a special character and first check whether a directory
named `ADMINI~1` exists under `Users`. Since no such directory actually
exists, the expansion fails.
Solution:
Treat `~` as a literal character in `do_path_expand`. Since the `~/`
case is already handled in `gen_expand_wildcards`, any remaining `~` is
just a literal character and will later be escaped to `\~` by
`file_pat_to_reg_pat` if needed.
(cherry picked from commit 7bf2ab4b87)
Problem:
When `vim.hl.range(0, …, { timeout = N })` is called, the deferred
`range_hl_clear` captures `buf=0`, which resolves to an arbitrary
"current buffer" at timeout. This may cause a stale highlight that never
gets cleared.
Solution:
Resolve `buf=0` explicitly, before `range_hl_clear` captures it.
(cherry picked from commit ec7dab077b)
Builds with -DNDEBUG do not contain the compilation string, so the test
will fail even if nvim is built with unibilium.
Similarly, zig builds only report "Compilation: TODO", so they also fail
the test even when build with unibilium
Detect both scenarios and skip the test, rather than failing it.
(cherry picked from commit b7fd8cbd2b)
Problem: tests: Test_backupskip() may read from $HOME
Solution: Set $HOME to an empty value, use --clean
(D Ben Knoble)
Even though we unset HOME, we can see via scriptnames that user files
are still sourced! One of my installed plugins warns when not compiled
with +python3, so this test has a "press Enter" prompt.
Use `--clean` like most other GetVimProg()'s do to fix it. Some tests
use `system()` instead, but that turns this test into a failure rather
than passing; I'm not sure why other tests don't suffer from this.
To prove to ourselves, we can use code like this:
diff --git i/src/testdir/test_options.vim w/src/testdir/test_options.vim
index a408e20e1..044364a54 100644
--- i/src/testdir/test_options.vim
+++ w/src/testdir/test_options.vim
@@ -1179,6 +1179,7 @@ func Test_backupskip()
" P_NODUP). Run this in a separate instance and write v:errors in a file,
" so that we see what happens on startup.
let after =<< trim [CODE]
+ call writefile([execute('scriptnames')], 'foo')
let bsklist = split(&backupskip, ',')
call assert_equal(uniq(copy(bsklist)), bsklist)
call writefile(['errors:'] + v:errors, 'Xtestout')
@@ -1196,7 +1197,7 @@ func Test_backupskip()
" unset $HOME, so that it won't try to read init files
let saveenv['HOME'] = getenv("HOME")
call setenv('HOME', v:null)
- exe 'silent !' . cmd
+ exe 'silent !' . cmd .. ' --cmd "echo &rtp"'
call assert_equal(['errors:'], readfile('Xtestout'))
" restore environment variables
Here, that causes "foo" to include a bunch of files under ~/.vim. I'm
not sure why this happens, but lets paper over it for the test.
We can also tell that (orthogonal to --clean) setting HOME='' works too.
Let's do that in addition since unsetting HOME isn't quite enough.
closes: vim/vim#200518d9c383aaf
Co-authored-by: D. Ben Knoble <ben.knoble+github@gmail.com>
(cherry picked from commit 0039a13fe4)
Problem:
Currently, only some filesystems (Btrfs, ext2, ext3, ext4) have full
support of accessing the `dirent` entry-type. On other filesystems,
`uv.fs_scandir_next` may return `nil` for an existing but unsupported
entry-type.
This means consumers (such as `fs.dir()`), cannot know if `nil` means
"non-existent" or "unsupported".
Solution:
Fall back to `uv.fs_lstat` when `etype` is `nil`; return "unknown" if it
fails.
(cherry picked from commit 4b5f026ac9)
Problem: quickfix: can set quickfixtextfunc in restricted/sandbox mode
(tacdm)
Solution: Disallow setting the quickfixtextfunc option from a sandbox
and restricted mode (Yegappan Lakshmanan).
closes: vim/vim#20305cb8510d470
Co-Authored-by: tacdm
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
(cherry picked from commit 1a064abb0a)
Problem: Cmdline ruler may be drawn for autocommand window.
Solution: Check that the current window is not an autocommand window
when deciding whether to draw the ruler.
(cherry picked from commit b58ce1ab79)
Problem: completion: cannot complete user cmd :K with 'ignorecase'
(rendcrx)
Solution: Skip the short-circuit when 'ignorecase' is set
(Yasuhiro Matsumoto)
The set_cmd_index() short-circuit for the :k command treats ":k<X>" as
":k {X}" (mark argument), which makes ":kz<Tab>" never reach the
command-name expansion path. With 'ignorecase' the same prefix on other
letters (":gz<Tab>") completes a user command like :Gz, so the result is
inconsistent. Skip the short-circuit when 'ignorecase' is set; default
behaviour is preserved so the existing :k tests still pass.
fixes: vim/vim#20241closes: vim/vim#20275b54e57ee54
Co-authored-by: Yasuhiro Matsumoto <mattn.jp@gmail.com>
(cherry picked from commit dee602e659)
Problem: If there are pending messages when starting to build the
runtime search path, a msg_show callback may invoke
runtime_search_path_validate() recursively.
Solution: Avoid msg_show callback by ensuring messages are flushed.
(cherry picked from commit 53da0c5060)
Problem:
Virtual lines above a line where a fold starts show `foldopen` in
`foldcolumn`.
Solution:
Check if the line below the virtual one is inside a fold that starts
higher up or if it's the start of a fold. In the latter case, don't show
anything in `foldcolumn` for the virtual line.
refactor: lint
(cherry picked from commit 526ae1cc1b)
Problem:
When mouse=n is set
- Dragging the mouse enters visual mode, and then stops listening for
mouse events.
- Double/Triple/Quad clicking performs selections.
- Clicking in visual mode moves the cursor (though not through the TUI).
Solution:
Explicitly gate mouse actions that affect visual mode with a check for
MOUSE_VISUAL. This matches the behavior described in :help mouse.
> If enabled for "v" (Visual mode) then double-click selects word-wise,
> triple-click makes it line-wise, and quadruple-click makes it
> rectangular block-wise.
(cherry picked from commit 24f7182390)
Problem: filetype: some html files are wrongly recognized as htmlangular
Solution: Use the \< atom to anchor ng-template and ng-content to start
of word (truffle)
Prevent false-positive htmlangular detection on words containing
'ng-template' or 'ng-content' as a substring (e.g. 'song-template',
'sing-content'). Anchor both branches with \< to require a word start,
matching the \<DTD\s\+XHTML\s idiom used five lines below.
related: neovim/neovim#39778.
closes: vim/vim#20246354ab1a69e
Co-authored-by: truffle <truffleagent@gmail.com>
(cherry picked from commit f3bb21e71d)
Problem:
Visual selection could end up in the wrong place after
nvim_buf_set_text or nvim_buf_set_lines. In some delete cases,
Visual.lnum was already clamped before the line shift happened, so the
adjustment got skipped.
Solution:
Split fix_cursor_cols into reusable fix_pos_col logic and reuse it
for Visual updates. Also adjust Visual.lnum before changed_lines so
the shift uses the original position before final clamping.
(cherry picked from commit 450ba41436)
Problem:
During startup, we manually trigger a useless and misleading `OptionSet`
event, which doesn't set `v:option_*` values (this is a limitation of
`nvim_exec_autocmds`).
ad4bc2d90c/runtime/lua/vim/_core/defaults.lua (L939).
Solution:
The `nvim_exec_autocmds('OptionSet',…)` call does not serve any purpose
since 5cbb9d613b, so just drop it.
Problem:
After closing and reopening Neovim, ]' and [' fail with E92: Buffer 0
not found for marks restored from ShaDa. Direct jumps like 'a work
because mark_get_local() rewrites fnum before returning, but ]' uses
getnextmark() which does not, leaving fnum = 0.
Solution:
Set .fnum = buf->b_fnum when restoring local marks from ShaDa.
(cherry picked from commit 0b96b3cd52)
Problem:
On "arm clang unittest" CI job, `make unittest` sometimes fails with
lots of these messages:
FAILED test/unit/testutil.lua @ 773: ...
test/unit/testutil.lua:773:
test/unit/testutil.lua:297: declaration specifier expected near '_Static_assert' at line 429
exit code: 256
stack traceback:
test/unit/testutil.lua:773: in function 'itp_parent'
test/unit/testutil.lua:811: in function <test/unit/testutil.lua:801>
Solution:
Update filter_complex_blocks.
(cherry picked from commit adb5d8a646)
Problem:
Using the `DiffTool` plugin (e.g. through `nvim -d ...` or `:DiffTool
<file1> <file2>` fails if a space is in one of the paths. This occurs
because the `diff` wraps the paths with quotes (`'`) if space
characters are present, which the line diff regex fails to parse.
Solution:
Update regex to handle quoted paths by matching the string within the
quotes, if it exists.
(cherry picked from commit 1e09b020e5)
Problem:
When using prompt_appendbuf with multi-element list,
the first item is concated and rest replace the prompt instead of
inserting the lines before the prompt.
Solution:
Concat first element with replace_buf and insert the rest of the list
with set_buffer_lines.
Signed-off-by: Tomas Slusny <slusnucky@gmail.com>
(cherry picked from commit a977e1077b)
Problem: 'title' is updated when changing the name of a non-current
buffer with nvim_buf_set_name().
Solution: Set RedrawingDisabled when renaming the buffer.
(cherry picked from commit 96fc7c150f)
Problem: Message redirection column for captured output is not reset
after :echon since (4260f73, e63346df).
Solution: Ensure msg_ext_append is set before the kind with :echon.
(cherry picked from commit ce9f4f0369)
Problem: The four pointer-resolution loops in u_read_undo() lack
an i != j guard, so a header whose uh_next.seq equals
its own uh_seq resolves uh_next.ptr to itself. On
buffer close, u_freeheader() sees uhp->uh_next.ptr !=
NULL and skips updating b_u_oldhead, so u_blockfree()
dereferences the freed header on the next iteration.
The same pattern applies to uh_prev, uh_alt_next and
uh_alt_prev. A crafted .un~ file in the same directory
as a text file can trigger the use-after-free and
subsequent double-free when the buffer is closed.
(Daniel Cervera)
Solution: Add an i != j guard to each of the four resolution
loops, matching the guard already present in the
duplicate-detection loop above.
closes: vim/vim#20168
Supported by AI
4f610f07b7
Co-authored-by: Christian Brabandt <cb@256bit.org>
(cherry picked from commit 2d5f56c0aa)
Problem: Crash with invalid shellredir/shellpipe value
(bfredl)
Solution: Validate the option and allow only a single "%s".
fixes: vim/vim#20157closes: vim/vim#2015984ae09dd79
Co-authored-by: Christian Brabandt <cb@256bit.org>
(cherry picked from commit ffe87d91f7)
Problem:
- Empty ranges have different `<`, `<=`, `has` and `intersect` semantics compared to regular ranges.
- `to_inclusive_pos` assumes that the end position of a range is exclusive, which is not true for empty ranges
Solution:
Special case empty ranges in these operations.
(cherry picked from commit 416f3482e7)
Problem:
When `:!` writes shell output to a buffer, write_output() splits on `\r`, `\n`,
and `\r\n`, replacing the terminator byte with NUL. For a binary-mode buffer
this is wrong: `\r` should be preserved verbatim, not treated as a line
terminator. This wrong behavior causes a file like `\r\n` round-trips through
`:%!cat` to `\n`.
This was masked when 'shelltemp' was enabled, because output went through a temp
file and the regular file I/O path handled binary-mode correctly. Switching the
default to 'noshelltemp' exposed the bug, since output is now piped directly
into write_output().
Solution:
In `write_output()`, skip the `\r` and `\r\n` splits for a binary-mode buffer;
only split on `\n`.
(cherry picked from commit 832a68835b)
Problem: select_spec tests a public function, but uses private enums.
Solution: Replace private enums with public enums.
(cherry picked from commit a61c8f3580)
feat(tui): restore 'ttyfast' to control tty requests
Problem:
When running nvim on a remote machine over SSH, if there is high ping,
then bg detection may not complete in time. This results in a warning
every time nvim is started. #38648
Solution:
Restore 'ttyfast' option and allow it to control whether or not bg
detection is performed. Because this is during startup and before any
user config or commands, we use the environment variable
`NVIM_NOTTYFAST` to allow disabling `ttyfast` during initialization.
Co-authored-by: Kyle <50718101+kylesower@users.noreply.github.com>
Problem: read_compound() in spellfile.c computes the size of the regex
pattern buffer using signed-int arithmetic on the attacker
controlled SN_COMPOUND sectionlen. With sectionlen=0x40000008
and UTF-8 encoding active the multiplication wraps to 27 while
the per-byte loop writes up to ~1B bytes, overflowing the heap.
Reachable when loading a crafted .spl file (e.g. via 'set spell'
after a modeline sets 'spelllang'). The cp/ap/crp allocations
have the same int + 1 overflow class (Daniel Cervera)
Solution: Use type size_t as buffer size and reject values larger than
COMPOUND_MAX_LEN (100000). Apply the same size_t treatment to
the cp/ap/crp allocations.
Github Advisory:
https://github.com/vim/vim/security/advisories/GHSA-q4jv-r9gj-6cwv9299332917
Co-authored-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
(cherry picked from commit 0976ce255b)
Problem:
parser_gc() calls ts_parser_delete() but leaves the userdata pointer
pointing to freed memory. If the GC finalizer runs at an unexpected time
(e.g. inside nvim_buf_get_lines #39411), a stale pointer could cause a crash.
Solution:
- NULL out `*ud` after ts_parser_delete() in parser_gc()
- Update parser_check() to handle NULL with a clear error message,
guarding all parser methods against UAF
Co-authored-by: Lewis Russell <lewis6991@gmail.com>
Signed-off-by: Szymon Wilczek <swilczek.lx@gmail.com>
(cherry picked from commit 0c3e6e1b0e)
