(AI-assisted)
Problem:
Zig build failing since 966e7a98f5:
FAILED ...r/work/neovim/neovim/test/functional/editor/xxd_spec.lua @ 22: xxd handles long lines in revert mode
Expected values to be equal.
Expected:
0
Actual:
134
stack traceback:
...r/work/neovim/neovim/test/functional/editor/xxd_spec.lua:26: in function <...r/work/neovim/neovim/test/functional/editor/xxd_spec.lua:22>
`huntype` parses the address column by repeatedly left-shifting
`want_off` (a signed `long`) by 4 bits per hex digit. After 16 hex
digits the value occupies all 64 bits; the next shift moves a non-zero
bit into the sign bit -- signed overflow, i.e. undefined behavior.
This happens on the Zig build because it enables UBSAN. The bug was
pre-existing, but 966e7a98f5 added a test that exercises it.
Solution
Shift through `unsigned long` and cast back to `long`, making the
overflow well-defined wrap-around.
Problem:
Backport commits may have subjects like "backport: fix(ui): ...".
(e.g. ecda67662f). git-cliff treats them as OTHER.
Solution:
Preprocess the subject to strip a leading "backport:" or "backport ".
Problem:
All (tested by me) terminals (xterm, st, ghostty, vte, foot, wezterm, konsole) do support `\x1b[0 q` as "reset cursor to default", but at least konsole and wezterm only understands `\x1b[0 q` as "reset to default", but have different behaviour on `\x1b[ q` (konsole sets "steady block", and wezterm does nothing (do not change cursor shape).
Solution:
Use `\x1b[0 q` would be more widely compatible "reset" sequence than `\x1b[ q`
P.S. actually, `xterm`, `ghostty` and `st` (with default config.h) sets "steady block" for both sequence, but still here `[0` behaves the same as `[`
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.
vim-patch:9.2.0626: Vim9: illegal characters allowed in dict key names with dot notation
vim-patch:9.2.0630: popup images: kitty images output in GUI mode
vim-patch:7bc4491ad CI: Bump the github-actions group across 1 directory with 3 updates
vim-patch:9.2.0634: GTK4: no minimum resize limit
vim-patch:ad8d21b14 runtime(syntax-tests): Rename test files on demand
vim-patch:8.1.2076: crash when trying to put a terminal in a popup window
vim-patch:8.2.1090: may use NULL pointer when skipping over name
vim-patch:8.2.1755: Vim9: crash when using invalid heredoc marker
vim-patch:8.2.1854: Vim9: crash when throwing exception for NULL string
vim-patch:8.2.1884: compiler warning for uninitialized variable
vim-patch:8.2.1924: Vim9: crash when indexing dict with NULL key
vim-patch:8.2.2292: Vim: expr test fails
vim-patch:8.2.3340: accessing uninitialized pointer
vim-patch:8.2.3359: Vim9: error for type when variable is not set
vim-patch:8.2.3396: when libcall() fails invalid pointer may be used
vim-patch:8.2.4000: Coverity warns for checking for NULL pointer after using it
vim-patch:8.2.4269: Coverity warns for using a NULL pointer
vim-patch:8.2.4539: when comparing special v:none and v:null are handled the same
vim-patch:8.2.4822: setting ufunc to NULL twice
vim-patch:9.0.1095: using freed memory when declaration fails
vim-patch:9.0.1375: crash when getting member of obj of unknown class
Problem: Keys valid in CTRL-X mode are never mapped while insert
completion is active, so <C-N> and <C-P> cannot be remapped
for completion started by complete().
Solution: Do not disable mappings in CTRL_X_EVAL mode. In this mode a
mapping cannot interfere with selecting the completion
method, which is what the no-mapping rule exists for.
related: vim/vim#6440
related: vim/vim#16880
closes: vim/vim#20489076585e6ad
Co-authored-by: Thomas M Kehrenberg <tmke8@posteo.net>
Problem: Deciding whether a group is in a "contains"/cluster list scans
the list and expands clusters on every check, which is slow for
syntaxes with large lists (e.g. plugins such as netrw).
Solution: Resolve each list once into a sorted, cluster-expanded set of
group IDs and use a binary search; cache it per syntax block and
drop the cache when syntax definitions change (Hirohito Higashi).
closes: vim/vim#2049048fbae4378
Co-authored-by: Hirohito Higashi <h.east.727@gmail.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Problem: The "%" command jumps to parens and braces inside comments,
unlike the "=" operator (cindent), which ignores them.
Solution: When 'comments' defines C-style comments and "%" is not in
'cpoptions', skip matching parens inside such comments, except
when the cursor is inside a comment so a match there can still
be found.
fixes: vim/vim#20329
related: vim/vim#20111
closes: vim/vim#20491b8a109dcfb
Co-authored-by: Hirohito Higashi <h.east.727@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Problem: gq (and other filters) on an empty buffer fail with
"E20: Mark not set": when the filter produces no output,
do_filter() still subtracts the line count from '[ and '],
pushing '] to line 0.
Solution: when the filter produces no output, put '[ and '] on a valid
line instead of subtracting past line 1 (glepnir).
related: neovim/neovim#30593
closes: vim/vim#19061aefbca2977
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.
Problem: Running `vim.filetype.match()` when current working directory
was removed from disk throws a `vim.fs.abspath` assertion error.
However, the matching might still be possible without trying to match
against full path (like if it is a known extension).
Solution: Safely compute absolute path and ignore it if it errors.
Problem: Clicking on a "below" virtual line interacts with the line
it is attached to for decor/drawing purposes, rather than
the line the extmark is placed at.
Solution: Account for "below" virtual lines when computing mouse line number.
Problem:
After 400f247397 the undotree test became flaky.
It seems to be caused by some kind of race condition where an entry in the undotree is not added when a change in the buffer happens (adding a bunch of sleep around buffer-change operations seems to fix the issue).
Solution:
Disable autoread plugin in undotree test.
Problem: Diagnostic virtual lines are hardcoded with "scroll" for
virt_lines_overflow option.
Solution: Add a `overflow` option to `virtual_lines` config
to support "wrap", "scroll", "trunc", and "auto".
Problem: Vim9: null value tests not sufficient
Solution: Add a more comprehensive test for null values
(Yegappan Lakshmanan)
closes: vim/vim#14701da9d345b3d
Co-authored-by: Yegappan Lakshmanan <yegappan@yahoo.com>
vim-patch:9.1.0053: MS-Windows: Key event test still fails
vim-patch:9.1.0057: MS-Windows: Key event test still fails
vim-patch:9.2.0540: tests: Test_mswin_event_mouse is flaky
vim-patch:9.2.0590: GTK4: drawing area loses focus shape on popup menu open
vim-patch:9.2.0595: MS-Windows: Wrong buffer size calculation for gvimext
vim-patch:9.2.0599: popup: title set with popup_setoptions() is not shown
vim-patch:057db5f7e CI: Bump actions/checkout in the github-actions group across 1 directory
vim-patch:48f2e78c8 runtime(doc): Add gtk4-slow doc tag
vim-patch:9e4042b2b translation(it): Update Italian man page
vim-patch:9.2.0604: tests: Test_mswin_event_mouse() is still flaky
vim-patch:9.2.0607: GTK4: inputdialog() does not work as expected
vim-patch:9.2.0613: tests: test_xxd_color2() checks for dash but uses sh
vim-patch:9.2.0616: GTK4: use-after-free on clipboard read timeout
vim-patch:9.2.0617: GvimExt: does not support different runtime dirs
vim-patch:9.2.0618: use-after-free in popup_getoptions() on dict_add() failure
vim-patch:9.2.0619: integer overflow in popup image size validation
vim-patch:0abffbff2 CI: Remove Cirrus CI and Coverity Scan
vim-patch:8.2.2583: Vim9: cannot compare result of getenv() with null
vim-patch:8.2.4087: cannot test items from an autoload script easily
vim-patch:8.2.4534: Vim9: "is" operator with empty string and null returns true
vim-patch:9.0.1674: help for builtin functions is not sorted properly
vim-patch:9.0.1727: minor problems with the teapot
vim-patch:9.0.1796: Vim9 problems with null_objects
vim-patch:9.0.1914: Vim9: few issues when accessing object members
vim-patch:9.0.2001: Vim9: segfault with islocked()
vim-patch:9.1.0385: Vim9: crash with null_class and null_object
vim-patch:9.1.0391: Vim9: could improve testing
vim-patch:9.1.0620: Vim9: segfauls with null objects
vim-patch:9.1.1865: tests: do not notice lines containing only a tab
vim-patch:9.1.2029: tests: the test_vim9_class.vim testfile is too long
vim-patch:9.2.0195: CI: test-suite gets killed for taking too long
Problem:
The 'autoread' option only checks for file changes reactively — on
FocusGained, :checktime, CmdlineEnter, etc. — by polling timestamps.
External changes are not detected until the user interacts with Neovim.
Solution:
Add a core module (runtime/lua/nvim/autoread.lua) enabled from
runtime/plugin/autoread.lua that watches each buffer's file using
vim._watch.watch() (libuv fs_event). On change detection it calls
:checktime, which invokes the existing buf_check_timestamp() logic
for reload/prompt handling. Watchers are managed via autocmds tied
to buffer lifecycle events and respect the 'autoread' option (global
and buffer-local).
Problem:
stdpath() may return a DOS 8.3 "shortened" filename, because Windows
truncates long usernames into `6ch~N` names in `TEMP/TMP` env vars.
We don't want to "leak" them into Nvim.
Solution:
For "run", pass `true` to `vim_FullName` to expand 8.3 filenames.
For "cache", call `os_realpath` to expand 8.3 filenames.
Co-authored-by: Justin M. Keyes <justinkz@gmail.com>
Problem:
Listing individual "skipped" notices for each non-actionable or
NVIM_TEST_INTEG test is noisy and makes the "skipped" list overwhelming
and less meaningful.
Solution:
- Mark non-actionable tests with "N/A".
- Report NVIM_TEST_INTEG tests as a separate 1-line summary.
Problem: possible integer overflow in spellfile tree bounds check
Solution: Rewrite the overflow check (Devon Krik)
The check 'startidx + len >= maxidx' uses signed int addition and can
overflow when startidx approaches INT_MAX. After overflow the wrapped
result bypasses the guard, allowing the subsequent loop to write
idxs[startidx + i] out of bounds on the heap.
Replace the addition with a safe subtractive check that maintains the
original >= semantics: len >= maxidx - startidx cannot overflow because
both operands are valid indices within [0, maxidx].
This fixes CWE-190 (Integer Overflow) leading to CWE-122 (Heap-based
Buffer Overflow).
closes: vim/vim#20483276920e138
Co-authored-by: Devon Kirk <hyder365@users.noreply.github.com>
Problem: 'autoindent' not stripped with virtualedit=onemore (after
9.2.0510).
Solution: Restore the decrement of cursor column when it's on NUL.
fixes: neovim/neovim#40183closes: vim/vim#204764b13277edd
- Explicitly match the dot accessor
- Exclude the qualifier when matching qualified function calls
The dot accessor lookbehind on builtin function calls was slow, matching
across expression based dictionary accessors was visually inconsistent,
and it's arguably more semantically correct.
closes: vim/vim#2048178094ff1d7
Co-authored-by: Doug Kearns <dougkearns@gmail.com>
Problem:
Flickering may occur when paging up/down in big files, as ranges for semantic
tokens are requested. This happens with LSP servers like gopls which return
"/full" semantic tokens if the file is too big, where we fall back to
viewport-range token retrievals.
Solution:
Broaden the requested ranges to one viewport of "overscan" on each side plus
some padding if possible:
(viewport_topline - viewport_height)..(viewport_botline + viewport_height)
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`.
Problem: tests: test_codestyle does not notice lines containing only a
tab
Solution: Fix the whitespace issue in eval.txt, update test_codestyle to
notice such issues (Hirohito Higashi)
closes: vim/vim#185955b5290ec02
Co-authored-by: Hirohito Higashi <h.east.727@gmail.com>
Problem:
`nvim -u NONE -d <(xxd one) <(xxd two)` has weird behavior.
Process substitution `<(...)` is a pipe and not a seekable file.
Test case:
cat /dev/random | head -c 10240 > one
cp one two
cat /dev/random | head -c 10240 >> two
nvim -u NONE -d <(xxd one) <(xxd two)
Solution:
Workaround the issue by skipping `nvim.difftool` if the 2 args are not
directories; fall-through to the builtin diff handling.
Signed-off-by: Tomas Slusny <slusnucky@gmail.com>
Problem: tests: test_cmdline.vim leaves swapfiles behind
Solution: Close open buffers using :bw! instead of :close!
bf1b41e387
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem: tests: test_sort.vim leaves swapfiles behind
Solution: Close open buffers using :bw! instead of :close!
f8c550fea0
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem: tests: test_registers fails when not run under X11
Solution: filter out warning message
b2a8df35e8
Co-authored-by: Christian Brabandt <cb@256bit.org>
Problem: unmarking a regular file when there are directories in the markings
list also removes the 2match highlight from those directories.
Solution: correctly rebuild the match pattern from the remaining markings,
using the right regex trailer for each entry in the list.
closes: vim/vim#20461affd4b5964
Co-authored-by: J. Paulo Seibt <jpseibt@gmail.com>
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.
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.
Problem:
Since 5181984d, `nvim_exec_autocmds({buf=…})` temporarily sets the
`buf`-window as current-window, and its statusline renders as
a "current-window statusline", regardless of the user's actual
current-window.
Solution:
Mark the statusline for redraw in `aucmd_restbuf()`, so it is redrawn
correctly after the event.
Co-authored-by: Luuk van Baal <luukvbaal@gmail.com>
Problem:
* 'shellcmdflag' states that its default value is set according to the
value of 'shell', but this behavior is not yet implemented on Windows.
The same applies to 'shellpipe', 'shellredir', and 'shellxquote'.
* On Windows, Git is often installed in paths containing spaces, and we
still do not correctly resolve the sh executable name as described in
'shell'.
* On Windows, the default value of 'shellslash' is always `false`,
which causes Unix-like shells to interpret `\` in paths returned by
some functions as escape charaters.
Solution:
Use a simple rule table to detect common shells (e.g. `cmd`,
`powershell`, shells whose names contain `csh` or `sh`) and apply
best-effort defaults, while leaving more complex scenarios to user
configuration.
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.
