When a UI detaches it will execute any UILeave events. Autocommands
cannot run in a libuv handler because they will in turn poll the event
loop, which results in recursive loop execution. So we schedule the
callback to detach the UI on the main event queue.
We also have to schedule the exit when the RPC channel is closed to
ensure it does not run until after `remote_ui_disconnect` has run,
otherwise it will hang.
Problem: Cannot disable individual captures and patterns in treesitter queries.
Solution:
* Expose the corresponding tree-sitter API functions for `TSQuery` object.
* Add documentation for `TSQuery`.
* Return the pattern ID from `get_captures_at_pos()` (and hence `:Inspect!`).
Problem: Unnecessary use of STRCAT() in au_event_disable(). STRCAT()
seeks to the end of new_ei, but here the end is already known.
Solution: Use STRCPY() and add p_ei_len to new_ei. Also fix a typo in a
comment. Add a test that 'eventignore' works in :argdo
(zeertzjq).
closes: vim/vim#16844969e11a18b
Cherry-pick p_ei_len from patch 9.1.0256.
The name of Cmdwin* events were changed to CmdWin* in 8ed2dbf6e2
without explanation. This commit changes them back.
This doesn't affect the creation or execution of autocommands. It only
affects the listing of autocommands.
Problem:
"use_float" in the if-condition causes a new buffer to be generated on every execution.
Solution:
- Remove the incorrect use_float check from the condition.
- Use buf_clear to properly clear the existing buffer.
This actually only affects the order in which Cmdline* and Cmdwin*
autocommands are listed, and it appears that the names of Cmdwin* were
changed to CmdWin* in 8ed2dbf6e2 without
explanation.
Also, remove the final NULL element from the names table.
Problem:
1. The original info text is truncated in pum_set_preview_text.
2. There is an extra newline character at the end of the info text.
3. Incorrect usage of plines_win leads to errors in calculating the window height.
Solution:
1. Improved string handling in pum_preview_set_text to safely process line content without modifying the original text. Now, the preview info is correctly preserved while splitting lines.
2. Do not append a trailing newline to the preview buffer when one is present at the end.
3. Set w_widher_inner in advance; otherwise, plines_win cannot correctly calculate the wrapped line height.
Problem:
"make lintdoc" is not validating vimdoc (:help) tags.
Solution:
- Call `lang_tree:parse()` to init the parser.
- Load netrw 🤢 explicitly, since it was moved to `pack/dist/opt/`.
- Fix invalid help tags.
This allows generated sources to be automatically rebuilt when modifying
hashy code.
Also, appending to NVIM_GENERATED_FOR_{HEADERS,SOURCES} in the middle of
custom commands is a bit strange. Move that after the custom commands.
Follow-up to #32768
This is slightly faster according to the benchmark.
This also makes it a build error if hashy is used incorrectly
(generating a case-insensitive hash function from mixed-case strings),
as duplicate case labels aren't allowed.
Problem: Unnecessary use of vim_tolower() in vim_strnicmp_asc().
Solution: Use TOLOWER_ASC() instead (zeertzjq).
It was passing *s1 and *s2 to vim_tolower(). When char is signed, which
is the case on most platforms, c < 0x80 is always true, so it already
behaves the same as TOLOWER_ASC().
closes: vim/vim#16826b7dc5d3b61
Use this function for hashy case-insensitive lookup, as it's ASCII-only.
Note that this function doesn't cast TOLOWER_ASC() argument to uint8_t,
so it'll treat a UTF-8 byte as smaller than NUL. It doesn't matter, as
one of the strings being compared is ASCII-only, and its behavior still
leads to consistent ordering.
It's no longer needed after #32152.
I'll not apply this change to Vim first, because Nvim's event name
aliases are a bit different from Vim.
After this change there is a small speedup in nvim_exec_autocmds as per
test/benchmark/autocmd_spec.lua.
Always allow the following four events to be nested, as they may contain
important information, and are triggered on the event loop, which may be
processed by a blocking call inside another autocommand.
- ChanInfo
- ChanOpen
- TermRequest
- TermResponse
There are some other events that are triggered on the event loop, but
they are mostly triggered by user actions in a UI client, and therefore
not very likely to happen during another autocommand, so leave them
unchanged for now.
Problem: too many strlen() calls in misc2.c
Solution: refactor misc2.c and use bsearch() instead of a linear search
to find matches in the key_names_table array (John Marriott).
This commit changes misc2.c to use bsearch() to perform string searches of
the key_names_table array.
Implementation detail:
- Some entries in this array have alternate names. Add field alt_name to
point to the alternate name.
- Some entries in this array are only available if a given feature is
defined. Keep them in the array, but add a boolean field enabled to
indicate if the record can be used or not. If the feature is not
available, the corresponding enabled field is set to FALSE.
In my measurements running the test suite on a huge non-gui build on
linux, the number of string comparisons in get_special_key_code():
Before (linear search): 2,214,957
After (binary search): 297,770
A side effect of this is 1477 calls to STRLEN() in
get_special_key_name() for the same test run are no longer necessary.
closes: vim/vim#167884a1e6dacbb
Skip the mouse shape changes.
Co-authored-by: John Marriott <basilisk@internode.on.net>
Problem:
When a function like vim.wait() is used, we continuously drain the main
event queue until it is empty, never stopping for user input. This means
the libuv timer never runs and the terminal never gets refreshed, so
emit_termrequest continously reschedules itself onto the same event
queue, causing an infinite loop.
Solution:
Use a separate "pending" event queue, where events that require a
terminal refresh are temporarily placed. Drain this queue after a
terminal refresh and events are copied back onto the main queue. This
prevents infinite loops since the main event queue will always be able
to properly drain.
Problem: Skipping over a concealed line for which `win_line()`
_should_ be called because it has `virt_lines_above = false`
lines associated with it.
Solution: Don't include such a line in `wl_lastlnum` from the line
above.
Problem: terminal mode cursor refresh logic has too many edge cases where it
fails when events change curbuf.
Solution: change the logic. Introduce cursor_visible to TerminalState to more
reliably track if terminal mode has changed busy. Move visibility handling to
refresh_cursor and move its call in refresh_terminal to terminal_check to avoid
temporarily changed curbufs from influencing cursor state.
This has the effect of "debouncing" shape/visibility updates to once per
terminal state tick (with the final attributes taking effect, as expected). I
think this is OK, but as a result it may also be warranted to update when
redrawing during the same state tick (e.g: from events executing :redraw); this
can be added later, if wanted.
Also move previous tests to a more appropriate place.
Problem: heap-buffer-overflow with 'nostartofline' and Ex command in
tag file.
Solution: Set cursor column when moving cursor to line 1 (zeertzjq).
closes: vim/vim#167963ed6659549
When a plugin registers a TermRequest handler there is currently no way
for the handler to know where the terminal's cursor position was when
the sequence was received. This is often useful information, e.g. for
OSC 133 sequences which are used to annotate shell prompts.
Modify the event data for the TermRequest autocommand to be a table
instead of just a string. The "sequence" field of the table contains the
sequence string and the "cursor" field contains the cursor
position when the sequence was received.
To maintain consistency between TermRequest and TermResponse (and to
future proof the latter), TermResponse's event data is also updated to
be a table with a "sequence" field.
BREAKING CHANGE: event data for TermRequest and TermResponse is now a
table
Problem: wildmenu highlighting in popup can be improved
Solution: Check if the completion items contain submatches of the
entered text (Girish Palya).
This update enables highlighting in the popup menu even when the matched
fragment or pattern appears within an item (string) rather than only at the
beginning. This is especially useful for custom completion, where menu items
may not always start with the typed pattern.
For specific use cases, refer to the two examples in
https://github.com/vim/vim/pull/16759
A sliding window approach is used with direct string comparison. Performance
is not a concern, as highlighting is applied only to displayed lines, even if
the menu list is arbitrarily long.
closes: vim/vim#167854ec46f3210
Co-authored-by: Girish Palya <girishji@gmail.com>
Problem: Last line in a window does not store correct `wl_lastlnum` if
lines below it are concealed (resulting in e.g. incorrect
cursor row).
Solution: Increment `wl_lastlnum` while it points to a line above a
concealed line.
Problem: using global variable for get_insert()/get_lambda_name()
(after v9.1.1151)
Solution: let the functions return a string_T object instead
(Yee Cheng Chin)
In vim/vim#16720, `get_insert()` was modified to store a string length in a
global variable to be queried immediately by another `get_insert_len()`
function, which is somewhat fragile. Instead, just have the function
itself return a `string_T` object instead. Also do the same for
`get_lambda_name()` which has similar issues.
closes: vim/vim#167750b5fe42071
Co-authored-by: Yee Cheng Chin <ychin.git@gmail.com>
Problem: mark '] wrong after copying text object (ubaldot)
Solution: Adjust position of '] for non-linewise, exclusive motions
(Jim Zhou)
related: vim/vim#16679closes: vim/vim#16772360a39ae6c
Co-authored-by: Jim Zhou <jimzhouzzy@gmail.com>
Problem: wrong flags passed down to nextwild()
(zeertzjq, after v9.1.1166)
Solution: only pass options flags (Girish Palya)
`options` needs to be passed into nextwild() since it may contain
WILD_KEEP_SOLE_ITEM which prevents the menu items list from getting
freed if there is only 1 item left (if `noselect` is set).
closes: vim/vim#16778d2219d547d
Co-authored-by: Girish Palya <girishji@gmail.com>
Problem: command-line auto-completion hard with wildmenu
Solution: implement "noselect" wildoption value (Girish Palya)
When `noselect` is present in `wildmode` and 'wildmenu' is enabled, the
completion menu appears without pre-selecting the first item.
This change makes it easier to implement command-line auto-completion,
where the menu dynamically appears as characters are typed, and `<Tab>`
can be used to manually select an item. This can be achieved by
leveraging the `CmdlineChanged` event to insert `wildchar(m)`,
triggering completion menu.
Without this change, auto-completion using the 'wildmenu' mechanism is
not feasible, as it automatically inserts the first match, preventing
dynamic selection.
The following Vimscript snippet demonstrates how to configure
auto-completion using `noselect`:
```vim
vim9script
set wim=noselect:lastused,full wop=pum wcm=<C-@> wmnu
autocmd CmdlineChanged : timer_start(0, function(CmdComplete, [getcmdline()]))
def CmdComplete(cur_cmdline: string, timer: number)
var [cmdline, curpos] = [getcmdline(), getcmdpos()]
if cur_cmdline ==# cmdline # Avoid completing each character in keymaps and pasted text
&& !pumvisible() && curpos == cmdline->len() + 1
if cmdline[curpos - 2] =~ '[\w*/:]' # Reduce noise by completing only selected characters
feedkeys("\<C-@>", "ti")
set eventignore+=CmdlineChanged # Suppress redundant completion attempts
timer_start(0, (_) => {
getcmdline()->substitute('\%x00$', '', '')->setcmdline() # Remove <C-@> if no completion items exist
set eventignore-=CmdlineChanged
})
endif
endif
enddef
```
fixes: vim/vim#16551closes: vim/vim#167592bacc3e5fb
Cherry-pick Wildmode_Tests() change from patch 9.0.0418.
Co-authored-by: Girish Palya <girishji@gmail.com>
Signed-off-by: Tomas Slusny <slusnucky@gmail.com>
Problem: Cannot use "g<" mapping with ext_messages. Mapping displays
the scrollback buffer since the last command, but the
scrollback buffer is not populated with ext_messages.
Solution: With ext_messages; store messages in the history that otherwise
wouldn't be. Mark them as temporary messages to be deleted when
the scrollback buffer would be cleared. To this end, make the
message history a doubly-linked list such that messages can be
removed from an arbitrary position.
Outlook: Default ext_messages UI might not show the hit-enter prompt
so we want "g<" to work as a recommended way to show messages
for the last command (prompted by an indicator).
Problem: Vim's diff block merging algorithm when doing a multi-file diff
is buggy when two different diff hunks overlap a single
existing diff block (after v9.1.0743)
Solution: fix a couple bugs in this logic:
1. Fix regression from v9.1.0743 where it's not correctly expanding the
2nd overlap correctly, where it always expands without taking into
account that this was always taken care of when the first overlap
happened. Instead, we should only grow the 2nd overlap if it overhangs
outside the existing diff block, and if we encounter a new overlapping
diff block (due to overlap chaining).
2. When we expand a diff block to match the hunk size on the orig side
(when handling the first overlap), we expand the same amount of lines
in the new side. This is not sound if there exists a second overlap
hunk that we haven't processed yet, and that hunk has different
number of lines in orig/new. Fix this by doing the corresponding
counter adjustment when handling 2nd/3rd/etc overlap by calculating
the difference in lines between orig and new side.
(Yee Cheng Chin)
closes: vim/vim#16768bc08ceb755
Co-authored-by: Yee Cheng Chin <ychin.git@gmail.com>
Problem: preinsert requires bot "menu" and "menuone" to be set,
but "menu" is redundant (after v9.1.1160)
Solution: preinsert only requires menuone (glepnir)
closes: vim/vim#1676394a045ed56
Co-authored-by: glepnir <glephunter@gmail.com>
Problem: The 'preinsert' feature requires Ctrl-Y to confirm insertion,
but Ctrl-Y only works when the popup menu (pum) is displayed.
Without enforcing this dependency, it could lead to confusing
behavior or non-functional features.
Solution: Modify ins_compl_has_preinsert() to check for both 'menu' and
'menuone' flags when 'preinsert' is set. Update documentation
to clarify this requirement. This avoids adding complex
conditional behaviors. (glepnir)
fixes: vim/vim#16728closes: vim/vim#16753a2c5559f29
Co-authored-by: glepnir <glephunter@gmail.com>
