Gary Johnson [Thu, 12 Jun 2025 19:15:07 +0000 (21:15 +0200)]
runtime(doc): remove an obsolete item from todo.txt
I cannot reproduce this defect. The replication instructions refer to
Mercurial version numbers and I no longer use Mercurial for my Vim
source. I tried comparing various Git versions of the example file,
src/if_cscope.c, and still couldn't reproduce the defect. Also, I use
Vim's diff mode a lot and haven't seen such a problem in a long time.
I am confident that it has been fixed.
Therefore, the item should be removed from todo.txt.
closes: #17534
Signed-off-by: Gary Johnson <garyjohn@spocom.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
jinyaoguo [Wed, 11 Jun 2025 19:30:01 +0000 (21:30 +0200)]
patch 9.1.1455: Haiku: dailog objects created with no reference
Problem: Haiku: dailog objects created with no reference
Solution: delete the objects before returning (jinyaoguo)
In the functions gui_mch_dialog() and gui_mch_font_dialog(), Dialog
objects are created but never escape the function scope. The call to
dialog->Go() only returns a boolean value and does not retain any
reference to the Dialog object itself, which may lead to potential
memory leak.
Fix this by deleting the object after using it.
closes: #17501
Signed-off-by: jinyaoguo <guo846@purdue.edu> Signed-off-by: Christian Brabandt <cb@256bit.org>
Ilya Grigoriev [Wed, 11 Jun 2025 19:07:35 +0000 (21:07 +0200)]
runtime(doc): add more pointers to 'completeopt'
Before this commit, I had trouble finding information about configuring
the insert mode completion. In particular, it was not clear that the
'wildopt' config that I already had in my vimrc does not apply here.
Also, `insert.txt` barely mentioned 'completeopt' except when
describing popups (I was more interested in bash-like behavior
where the unique prefix of all completions is completed first).
I'm hoping these edits will make the relevant docs easier to find.
closes: #17515
Signed-off-by: Ilya Grigoriev <ilyagr@users.noreply.github.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Doug Kearns [Tue, 10 Jun 2025 19:15:22 +0000 (21:15 +0200)]
runtime(filetype): Use s:StarSetf for all patterns ending in *
Problem: Some filetype autocmds with patterns ending in * do not skip
filenames matching g:ignored_patterns.
Solution: Move these autocmds to the appropriate section and call
s:StarSetf() to set the filetype.
- Affected filetypes: dosini, execline, foam, messages, nginx, tmux.
- Convert foam filetype patterns to use wildcard matching.
closes: #17422
Signed-off-by: Doug Kearns <dougkearns@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
zeertzjq [Tue, 10 Jun 2025 18:31:44 +0000 (20:31 +0200)]
patch 9.1.1450: Session has wrong arglist with :tcd and :arglocal
Problem: Session has wrong arglist with :tcd and :arglocal.
Solution: Also use absolute path for :argadd when there is tabpage-local
directory (zeertzjq).
related: neovim/neovim#34405
closes: #17503
Signed-off-by: zeertzjq <zeertzjq@outlook.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Girish Palya [Mon, 9 Jun 2025 18:43:03 +0000 (20:43 +0200)]
patch 9.1.1445: negative matchfuzzy scores although there is a match
Problem: negative matchfuzzy scores although there is a match
(Maxim Kim)
Solution: reset the score if a match has been found but the score is
negative (Girish Palya)
The fuzzy algorithm may miss some matches in long strings due to recursion
limits. As a result, the score can end up negative even when matches exist.
In such cases, reset the score to ensure it is non-negative.
fixes: ##17449
closes: #17469
Signed-off-by: Girish Palya <girishji@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
jinyaoguo [Mon, 9 Jun 2025 18:31:17 +0000 (20:31 +0200)]
patch 9.1.1443: potential buffer underflow in insertchar()
Problem: potential buffer underflow in insertchar()
Solution: verify that end_len is larger than zero
(jinyaoguo)
When parsing the end-comment leader, end_len can be zero if
copy_option_part() writes no characters. The existing check
unconditionally accessed lead_end[end_len-1], causing potential
underflow when end_len == 0.
This change adds an end_len > 0 guard to ensure we only index lead_end
if there is at least one character.
closes: #17476
Signed-off-by: jinyaoguo <guo846@purdue.edu> Signed-off-by: Christian Brabandt <cb@256bit.org>
Gary Johnson [Mon, 9 Jun 2025 18:19:35 +0000 (20:19 +0200)]
patch 9.1.1442: tests: Test_diff_fold_redraw() is insufficient
Problem: tests: Test_diff_fold_redraw() is insufficient
(after v9.1.1439, Christ van Willegen)
Solution: improve the test (Gary Johnson)
The original Test_diff_fold_redraw() function, added 2025-06-08 at patch
9.1.1439, had a bug and didn't do a very good job of testing the fold
behavior. This new version is simpler and more thorough.
The bug was that it checked the fold state of one window twice instead
of checking both windows.
closes: #17492
Signed-off-by: Gary Johnson <garyjohn@spocom.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Girish Palya [Sun, 8 Jun 2025 14:20:06 +0000 (16:20 +0200)]
patch 9.1.1441: completion: code can be improved
Problem: completion: code can be improved
Solution: remove reposition_match() and use mergesort_list(),
for fuzzy completion, sort by fuzzy score immediately after
setting a new leader (Girish Palya)
John Marriott [Sun, 8 Jun 2025 14:05:53 +0000 (16:05 +0200)]
patch 9.1.1440: too many strlen() calls in os_win32.c
Problem: too many strlen() calls in os_win32.c
Solution: refactor code and remove calls to strlen()
(John Marriott)
The following changes have been made:
- In mch_init_g():
- refactor to remove calls to STRLEN().
- use vim_strnsave() instead of vim_strsave().
- set a flag if vimrun_path is stored in allocated memory so it can be
freed at exit.
- In mch_exit() free vimrun_path if it was stored in allocated memory.
- In fname_case() make a small optimisation by measuring the length of
name only if needed.
- In copy_extattr() make a small optimisation by replacing call to
STRCAT() with STRCPY().
closes: #17462
Signed-off-by: John Marriott <basilisk@internode.on.net> Signed-off-by: Christian Brabandt <cb@256bit.org>
Gary Johnson [Sun, 8 Jun 2025 14:00:06 +0000 (16:00 +0200)]
patch 9.1.1439: Last diff folds not merged
Problem: Last diff folds not merged (after v8.1.1922)
Solution: loop over all windows in the current tabpage and update all
folds (Gary Johnson)
This commit fixes a bug where the last two folds of a diff are not
merged when the last difference between the two diff'd buffers is
resolved.
Normally, when two buffers are diff'd, folding is used to show only the
text that differs and to hide the text that is the same between the two
buffers. When a difference is resolved by making a block of text the
same in both buffers, the folds are updated to merge that block with the
folds above and below it into one closed fold.
That updating of the folds did not occur when the block of text was the
last diff block in the buffers.
The bug was introduced by this patch on August 24, 2019:
patch 8.1.1922: in diff mode global operations can be very slow
Problem: In diff mode global operations can be very slow.
Solution: Do not call diff_redraw() many times, call it once when
redrawing. And also don't update folds multiple times.
Unfortunately, folds were then not updated often enough.
The problem was fixed by adding a short loop to the ex_diffgetput()
function in diff.c to update all the folds in the current tab when the
last difference is removed.
A test for this was added to test_diffmode.vim. Two of the reference
screen dumps for another test in that file,
Test_diffget_diffput_linematch(), had to be changed to have all the
folds closed rather than to have the last diff block remain open.
closes: #17457
Signed-off-by: Gary Johnson <garyjohn@spocom.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Problem: tests: Test_breakindent_list_split() fails
(Phạm Bình An)
Solution: Always reset "&columns" and "&lines" for GUI builds
(Aliaksei Budavei)
Ensure that "&columns" and "&lines" are always set to their
default values before calling "SetUp()", if any, for EACH
test run by a GUI build to avoid yet-to-be-run tests from
inheriting possibly changed values (after window resizing)
and leading to broken assumptions about available estate and
occasional test failures.
fixes: #17453
closes: #17447
Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
runtime(vim): vimHLGroup is not highlighted correctly
Problem: vimHLGroup is not highlighted in "hi def link"
and "hi clear" commands
Solution: highlight vimHLGroup similarly to vimGroup
(Eisuke Kawashima)
Phạm Bình An [Thu, 5 Jun 2025 19:21:35 +0000 (21:21 +0200)]
runtime(typescript): remove Fixedgq() function from indent script
Problem:
1. The `Fixedgq()` function is broken (see #17412)
2. The `'formatexpr'` for Typescript is not documented, which causes
confusion to users when they try to set `'formatprg'`, since
`'formatexpr'` always takes precedence over `'formatprg'`. See also
https://github.com/HerringtonDarkholme/yats.vim/issues/209
3. Typescript already has a very good and popular formatter called
`prettier`, that can be easily integrated to Vim via `'formatprg'`
(see #16989). I don't think there are any good reasons to reinvent a
half-baked version in Vim.
Solution: Remove the Fixedgq() 'formatexpr' function.
fixes: #17412
closes: #17452
Signed-off-by: Phạm Bình An <phambinhanctb2004@gmail.com> Signed-off-by: Doug Kearns <dougkearns@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Girish Palya [Thu, 5 Jun 2025 19:04:29 +0000 (21:04 +0200)]
patch 9.1.1435: completion: various flaws in fuzzy completion
Problem: completion: various flaws in fuzzy completion
Solution: fix the issues (Girish Palya)
- Remove the brittle `qsort()` on `compl_match_array`.
- Add a stable, non-recursive `mergesort` for the internal doubly
linked list of matches.
- The sort now happens directly on the internal representation (`compl_T`),
preserving sync with external structures and making sorting stable.
- Update fuzzy match logic to enforce `max_matches` limits after
sorting.
- Remove `trim_compl_match_array()`, which is no longer necessary.
- Fixe test failures by correctly setting `selected` index and
maintaining match consistency.
- Introduce `mergesort_list()` in `misc2.c`, which operates generically
over doubly linked lists.
- Remove `pum_score` and `pum_idx` variables
fixes: #17387
closes: #17430
Signed-off-by: Girish Palya <girishji@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
glepnir [Mon, 2 Jun 2025 17:45:41 +0000 (19:45 +0200)]
patch 9.1.1426: completion: register contents not completed
Problem: CTRL-X CTRL-R only completes individual words from registers,
making it difficult to insert complete register content.
Solution: Add consecutive CTRL-X CTRL-R support - first press completes
words, second press completes full register lines, similar to
CTRL-X CTRL-L and CTRL-X CTRL-P behavior (glepnir).
closes: #17395
Signed-off-by: glepnir <glephunter@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
patch 9.1.1425: tabpanel: there are still some problems with the tabpanel
Problem: tabpanel: there are still some problems with the tabpanel with
column handling
Solution: fix the problems and refactor Tabpanel feature (Hirohito
Higashi).
fixes: #17423
fixes: #17332
closes: #17336
Signed-off-by: Hirohito Higashi <h.east.727@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Girish Palya [Sun, 1 Jun 2025 18:11:59 +0000 (20:11 +0200)]
patch 9.1.1424: PMenu selection broken with multi-line selection and limits
Problem: PMenu selection broken with multi-line selection and limits
(Maxim Kim)
Solution: update completion match index when limiting the completion
sources (Girish Palya)
fixes: #17394
closes: #17401
Signed-off-by: Girish Palya <girishji@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
patch 9.1.1423: :tag command not working correctly using Vim9 Script
Problem: :tag command not working correctly using Vim9 Script
Solution: inject a ':' before the numeric address, to make the command
valid in Vim9 context
Girish Palya [Sun, 1 Jun 2025 17:40:00 +0000 (19:40 +0200)]
patch 9.1.1422: scheduling of complete function can be improved
Problem: scheduling of complete function can be improved
Solution: call user completion functions earlier when just determining
the insertion column (Girish Palya)
This change improves the scheduling behavior of async user-defined
completion functions (such as `F{func}`, `F`, or `'o'` values in the
`'complete'` option), particularly benefiting LSP clients.
Currently, these user functions are invoked twice:
1. First with `findstart = 1` to determine the completion start
position.
2. Then with `findstart = 0` to retrieve the actual matches.
Previously, both calls were executed back-to-back. With this change, the
first call (`findstart = 1`) is performed earlier—before any matches are
gathered from other sources.
This adjustment gives event-driven completion sources (e.g., LSP
clients) more time to send their requests while Vim concurrently
collects matches from other sources like the current buffer.
Not sure about the real-world performance gains, but this approach
should, in theory, improve responsiveness and reduce latency for
asynchronous completions.
Phạm Bình An [Sun, 1 Jun 2025 16:58:09 +0000 (18:58 +0200)]
patch 9.1.1421: tests: need a test for the new-style tutor.tutor
Problem: tests: need a test for the new-style tutor.tutor, patch
9.1.1384 broke the expected positions for the signs
Solution: Update all number keys in tutor.tutor.json to match the
correct line numbers in tutor.tutor, replace tabs by spaces,
add a screen-dump test to verify it does not regress
(Pham Bình An)
closes: #17416
Signed-off-by: Phạm Bình An <phambinhanctb2004@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
runtime(java): Match annotation- and interface-type names of "java.lang"
Complement the documented support for the recognition of all
public types of the "java.lang" package (":help java.vim").
(The original syntax item generator may have, inadvertently,
contributed via suppressing "NullPointerException"s to not
having annotation and interface types qualify in general.)
Also, re-link usage instructions for the alternative syntax
item generator to a rolling "master"'s version.
closes: #17419
Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Luuk van Baal [Sat, 31 May 2025 10:10:31 +0000 (12:10 +0200)]
patch 9.1.1419: It is difficult to ignore all but some events
Problem: It is difficult to ignore all but some events.
Solution: Add support for a "-" prefix syntax in '(win)eventignore' that
subtracts an event from the ignored set if present
(Luuk van Baal).
closes: #17392
Signed-off-by: Luuk van Baal <luukvbaal@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
- Match comments and trailing bar after :set without args.
- Match the <...> form for key code options.
- Remove orphaned vim_ex_python[3x]* dump files (Aliaksei Budavei).
closes: #17397
Signed-off-by: Doug Kearns <dougkearns@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
Drew Vogel [Wed, 28 May 2025 19:13:52 +0000 (21:13 +0200)]
patch 9.1.1418: configures GUI auto detection favors GTK2
Problem: configures GUI auto detection favors GTK2
Solution: make configure favor GTK3 over GTK2 for the GUI
when auto detecting the gui toolkit (Drew Vogel).
Prior to these changes if the dev packages for both GTK2 and GTK3 were
installed, the `--enable-gui=auto` would used GTK2. After these changes
it will use GTK3. Users can still use `--enable-gui=gtk2` to
specifically select GTK2.
In addition to the prioritization change, this also brings some cleanups
to the GTK autoconf code:
* The `AM_PATH_GTK` macro had an unused third argument that has been
removed.
* The `AM_PATH_GTK` macro checked the `SKIP_GTK2` & `SKIP_GTK3`
variables but the code that decided whether to call it also checked
those. Now just the calling code does so.
* The `AM_PATH_GTK` macro set a default minimum version based on
`SKIP_GTK2` and `SKIP_GTK3` but the calling code was also expected to
pass a version. Now the calling code _must_ pass a version.
* The GTK test program previous used `gtk_(major|minor|micro)_version`
as all of: a C variable name, a C macro provided only by GTK2, and an
autoconf variable name. It also needlessly parsed a `x.y.z` version
string when the same string was already parsed by autoconf + sed. Now
the parsed values are used directly in the test program.
* The GTK test program previous created a test program `conf.gtktest`
which was cleaned up by the autoconf script. This appeared to be a
crude way to debug whether an erroring configure run had actually run
the test program. Instead the autoconf script now outputs more messaging
and the user can check `config.log` to determine the status of the
configure script.
I'm not an autoconf expert and I don't have access to some of the older
systems we try to support with gvim. So I would very much appreciate if
anyone could run this on their systems to ensure it doesn't misbehave.
While my motivation here is mainly to further establish GTK3 as the
primary GUI mode, this should at least partially address the concern
described in #15437.
Here are a few test runs with both GTK 2 and GTK 3 installed:
```
checking --enable-gui argument... yes/auto - automatic GUI support
checking whether or not to look for GTK+ 2... yes
checking whether or not to look for GNOME... no
checking whether or not to look for GTK+ 3... no
checking whether or not to look for Motif... yes
checking for pkg-config... /usr/bin/pkg-config
checking --disable-gtktest argument... gtk test enabled
checking for pkg-config gtk+-2.0... found
checking for GTK - version >= 2.2.0... yes; found version 2.24.33
checking ability to compile GTK test program... yes
```
```
checking --enable-gui argument... yes/auto - automatic GUI support
checking whether or not to look for GTK+ 2... no
checking whether or not to look for GTK+ 3... yes
checking whether or not to look for Motif... yes
checking for pkg-config... /usr/bin/pkg-config
checking --disable-gtktest argument... gtk test enabled
checking for pkg-config gtk+-3.0... found
checking for GTK - version >= 3.0.0... yes; found version 3.24.49
checking ability to compile GTK test program... yes
```
```
--with-features=huge \
```
```
checking --enable-gui argument... yes/auto - automatic GUI support
checking whether or not to look for GTK+ 2... yes
checking whether or not to look for GNOME... no
checking whether or not to look for GTK+ 3... yes
checking whether or not to look for Motif... yes
checking for pkg-config... /usr/bin/pkg-config
checking --disable-gtktest argument... gtk test enabled
checking for pkg-config gtk+-3.0... found
checking for GTK - version >= 3.0.0... yes; found version 3.24.49
checking ability to compile GTK test program... yes
```
```
checking --enable-gui argument... yes/auto - automatic GUI support
checking whether or not to look for GTK+ 2... yes
checking whether or not to look for GNOME... no
checking whether or not to look for GTK+ 3... yes
checking whether or not to look for Motif... yes
checking for pkg-config... /usr/bin/pkg-config
checking --disable-gtktest argument... gtk test disabled
checking for pkg-config gtk+-3.0... found
checking for GTK - version >= 3.0.0... yes; found version 3.24.49
```
```
checking --enable-gui argument... GTK+ 2.x GUI support
checking for pkg-config... /usr/bin/pkg-config
checking --disable-gtktest argument... gtk test enabled
checking for pkg-config gtk+-2.0... found
checking for GTK - version >= 2.2.0... yes; found version 2.24.33
checking ability to compile GTK test program... yes
```
```
checking --enable-gui argument... GTK+ 3.x GUI support
checking for pkg-config... /usr/bin/pkg-config
checking --disable-gtktest argument... gtk test enabled
checking for pkg-config gtk+-3.0... found
checking for GTK - version >= 3.0.0... yes; found version 3.24.49
checking ability to compile GTK test program... yes
```
And here is a similar run with the GTK 3 dev package removed:
glepnir [Wed, 28 May 2025 18:39:34 +0000 (20:39 +0200)]
patch 9.1.1417: missing info about register completion in complete_info()
Problem: missing info about register completion in complete_info()
(after v9.1.1408)
Solution: update documentation and mention that register is used as
source, add a test (glepnir)
closes: #17389
Signed-off-by: glepnir <glephunter@gmail.com> Signed-off-by: Christian Brabandt <cb@256bit.org>
patch 9.1.1415: potential use-after free when there is an error in 'tabpanel'
Problem: potential use-after free when there is an error in 'tabpanel'
option (@char101, after v9.1.1391)
Solution: check if p_tpl has been set to null before accessing it again.
While at it slightly change starts_with_percent_and_bang() and use the
existing opt_name and opt_scope variables.
projects / thirdparty / vim.git / log
