Windows: In order for jobstart(['foo']), system(['foo']) to find
"foo.cmd", we must replace "foo" with "foo.cmd" before sending `argv` to
process_spawn().
Rationale: jobstart([…]), system([…]) "executable" semantics should be
consistent with the VimL executable() function.
fix#9569
related: #10554
Typically most shell output is the result of non-trivial work, so it
would not blast stdout instantaneously. To more closely simulate that
typical scenario, change `shell-test REP` to wait 1 millisecond between
iterations.
Factor `get_snapshot()` out of `print_snapshot()`, so that callers can
get a table (for use with `expect()`) instead of the string form.
Try to use this to fix indeterminism in `searchhl_spec.lua`.
- Since the screen state is collected by `screen:expect_unchanged()`,
we don't need a deterministic initial state (which would then be
hardcoded into the test). This allows us to check "did anything
change in the last N ms?" rather than "did anything change compared
to a hardcoded screen-state?"
- This may end up fruitless, because `expect_unchanged()` depends on
timing to wait for an initial "current state".
For "backwards" duration, reltimefloat() should return negative value
like its counterpart reltimestr().
ref bab24a88ab
ref 06af88cd72
ref #10521fix#10452
It would previously fail with `set shell=sh` (no slash).
For the test itself we can just use a non-existing (fake) shell, because
it is only about the verbose output.
Ref: https://github.com/neovim/neovim/issues/9330
Also fix V576: use width specification
> Incorrect format. Consider checking the third actual argument of the
> 'sscanf' function. It's dangerous to use string specifier without width
> specification. Buffer overflow is possible.
- reltimestr(): Produce negative value by comparing the unsigned
proftime_T value to INT64_MAX.
https://github.com/neovim/neovim/issues/10452#issuecomment-511155132
1. The interfaces of nearly all platforms return uint64_t. INT64_MAX is
only half of that.
2. Low-level interfaces like this typically define that there is no
fixed starting point. The only guarantees are that it's (a)
monotonically increasing at a rate that (b) matches real time.
ref 06af88cd72fix#10452
* Fix problem that 1byte extra memory was allocated in is_executable_in_path
* Revert "Revert "tests: executable_spec: enable pending test #10443" (#10454)"
This reverts commit 13fbeda0e5.
The interaction between 'winblend' and doublewidth chars in the background
does not look very good. But check no chars get incorrectly placed
at least.
Also check that hidden EndOfBuffer region (from style="minimal") blends
correctly.
- Move __gcov_flush to process_spawn, for more reliable coverage
tracking of subprocesses
- Travis: use GCOV_ERROR_FILE
- codecov: use "-X fix" to skip "fixing" uploaded coverage data; it
should be handled by codecov's backend instead.
- AppVeyor: no $PATH mangling, which breaks with the improved coverage tracking
due to missing .dll in PATH.
Note: test doesn't fail on master. I cannot reproduce the glitches with
-u NONE, probably it requires interfering events. But add some coverage
for these checks at least.
ext_message doesn't set msg_col. Add a space and let client deal with
wrapping. When using silent redirect show the unwrapped message form.
Removed check is already part of msg_advance()
Flaky test (osx):
[ FAILED ] ...is/build/neovim/neovim/test/functional/lua/loop_spec.lua @ 23: vim.loop timer
...is/build/neovim/neovim/test/functional/lua/loop_spec.lua:56: Expected objects to be the same.
Passed in:
(number) 0
Expected:
(number) 2
stack traceback:
...is/build/neovim/neovim/test/functional/lua/loop_spec.lua:56: in function <...is/build/neovim/neovim/test/functional/lua/loop_spec.lua:23>
It was bumped from sleeping for 20ms to 50ms in d33aaa0f5f already.
[ ERROR ]...neovim/neovim/test/functional/autocmd/termclose_spec.lua @ TermClose event triggers when fast-exiting terminal job stops
test/functional/helpers.lua:96: Vim(call):E900: Invalid channel id
stack traceback:
test/functional/helpers.lua:96: in function 'request'
test/functional/helpers.lua:254: in function 'command'
...neovim/neovim/test/functional/autocmd/termclose_spec.lua:23: in function <...neovim/neovim/test/functional/autocmd/termclose_spec.lua:20>
Rationale: the purpose of nvim_execute_lua is to simply call lua code with lua
values. If a lua function expects a floating point value, it should be enough
to specify a float as argument to nvim_execute_lua.
However, make sure to preserve the existing roundtripping behavior of
API values when using `vim.api` functions. This is covered by existing
lua/api_spec.lua tests.
* build: update some test dependencies
* luacheck ignores
* BuildLua: add ${BUSTED} to depends for ${BUSTED_LUA}
This is required to rebuild it when busted gets updated.
This matches Vim behavior. From `:help :ls` :
R a terminal buffer with a running job
F a terminal buffer with a finished job
? a terminal buffer without a job: `:terminal NONE`
TODO: implement `:terminal NONE`.
ref #10349
Seen on Travis (osx):
[ RUN ] timers can be stopped from the handler: FAIL
.../build/neovim/neovim/test/functional/eval/timer_spec.lua:167: Expected objects to be the same.
Passed in:
(number) 2
Expected:
(number) 3
stack traceback:
.../build/neovim/neovim/test/functional/eval/timer_spec.lua:167: in function <.../build/neovim/neovim/test/functional/eval/timer_spec.lua:153>
<Paste> is a 3-byte sequence and the beginning one or two bytes can appear at
the very end of the typeahead buffer. When this happens, we were exiting from
`vgetorpeek()` instead of reading more characters to see the complete sequence.
I think this should fix#7994 -- at least partially. Before this change, when I
paste exactly 64 characters into a freshly booted instance, I get what I pasted
plus the literal text "<Paste>" at the end. Nvim also stays in nopaste mode.
The attached test case fails in this manner without the code change.
Fix#7994
* screen: Fix to draw signs with combining characters.
The buffer size for signs can be too small, because the upper length
limit of a sign can be 56 bytes. If combining characters are only two
bytes in size, this reduces to 32 bytes.
* screen: Adjust buffer size to maximal sign column count
There might be an existing job already - maybe due to some other test,
but in this case there was only one failure in the test run.
```
[----------] Running tests from C:/projects/neovim/test/functional\api\proc_spec.lua
[ RUN ] api nvim_get_proc_children returns child process ids: ERR
test\functional\helpers.lua:392:
retry() attempts: 450
C:/projects/neovim/test/functional\api\proc_spec.lua:22: Expected objects to be the same.
Passed in:
(number) 2
Expected:
(number) 1
stack traceback:
test\functional\helpers.lua:392: in function 'retry'
C:/projects/neovim/test/functional\api\proc_spec.lua:21: in function <C:/projects/neovim/test/functional\api\proc_spec.lua:17>
```
https://ci.appveyor.com/project/neovim/neovim/builds/25461215/job/8ns204v6091iy9rs?fullLog=true#L2672
* ci: AppVeyor: set GCOV_ERROR_FILE
This prevents the warnings/errors to be spilled into test results,
causing them to fail them, e.g.:
[ FAILED ] C:/projects/neovim/test/functional\core\main_spec.lua @ 97: Command-line option -s errors out when trying to use nonexistent file with -s
C:/projects/neovim/test/functional\core\main_spec.lua:98: Expected objects to be the same.
Passed in:
(string) 'Cannot open for reading: "Xtest-functional-core-main-s.nonexistent": no such file or directory
profiling:C:\projects\neovim\build/src/nvim/CMakeFiles/nvim.dir/buffer.c.gcda:Data file mismatch - some data files may have been concurrently updated without locking support
'
Expected:
(string) 'Cannot open for reading: "Xtest-functional-core-main-s.nonexistent": no such file or directory
'
stack traceback:
C:/projects/neovim/test/functional\core\main_spec.lua:98: in function <C:/projects/neovim/test/functional\core\main_spec.lua:97>
For reference, the locking appears to have been reworked for gcc 9.1 [1].
1: https://github.com/gcc-mirror/gcc/commit/56621355b
helpers.clear: keep GCOV_ERROR_FILE in environment
* ci: AppVeyor: remove MINGW_64 config (used with cov now)
Also:
- run MINGW_64-gcov first, and with PRs, since it provides coverage.
This might be required on (slower) CI.
[ RUN ] timers doesn't mess up the cmdline: ERR
test/functional/ui/screen.lua:562: expected intermediate screen state before final screen state
stack traceback:
test/functional/ui/screen.lua:562: in function '_wait'
test/functional/ui/screen.lua:366: in function 'expect'
.../build/neovim/neovim/test/functional/eval/timer_spec.lua:221: in function <.../build/neovim/neovim/test/functional/eval/timer_spec.lua:199>
Ref: https://travis-ci.org/neovim/neovim/jobs/544974506#L3861
Problem: When we changed startup to wait for the TUI (like a remote UI),
we forgot to set os/input.c:global_fd. That used to be done by
input_start().
Solution: Initialize os/input.c:global_fd before initializing libtermkey
(termkey_new_abstract) so that tui_get_stty_erase() and
friends can inspect the correct fd.
fixes#10134close#10174
Problem: Sign message not translated and inconsistent spacing.
Solution: Add _() for translation. Add a space. (Ken Takata) Also use
MSG_BUF_LEN instead of BUFSIZ.
d730c8e297
Problem: Placing signs can be complicated.
Solution: Add functions for defining and placing signs. Introduce a group
name to avoid different plugins using the same signs. (Yegappan
Lakshmanan, closesvim/vim#3652)
162b71479b
Unfortunately we cannot indiscriminately replace xfree() with
XFREE_CLEAR(), because comparing pointers after freeing them is a common
pattern. Example in `tv_list_remove_items()`:
xfree(li);
if (li == item2) {
break;
}
Instead we can do it selectively/explicitly.
ref #1375
The test.functional.helpers and test.unit.helpers modules now include
all of the public functions from test.helpers, so there is no need to
separately require('test.helpers').
This is where "pure functions" can live, which can be shared by Nvim and
test logic which may not have a running Nvim instance available.
If in the future we use Nvim itself as the Lua engine for tests, then
these functions could be moved directly onto the `vim` Lua module.
closes#6580
Automatically include all "global helper" util functions in the
unit.helpers and functional.helpers and modules. So tests don't need to
expicitly do:
local global_helpers = require('test.helpers')
Previously, ordinary redraws were missing from terminal mode. Instead,
there was an async callback that invoked update_screen() on terminal
data regardless of mode (as if :redraw! was invoked by a timer).
This created some issues:
- async changes to an unrelated ordinary buffer were not always redrawn in
terminal mode
- screen cursor position was not properly updated in terminal mode (partial
fix, will be properly fixed in a follow up PR)
- ad-hoc logic was needed for interaction with special states such as
inccommand or horizontal wildmenu.
Instead redraw terminal mode just like any other state. This disables forced
redraws in cmdline mode, which were inconisent which async changes to
normal buffers (which are not redrawn in cmdline mode).
- redraw! in an invisible buffer rendered the screen unusable.
- storing the autocmd window handle and using it in API function could lead
to crashes. Unregister the handle when the window is not active.
closes#9136
- Treat empty {rhs} like <Nop>
- getchar.c: Pull "repl. MapArg termcodes" into func
The "preprocessing code" surrounding the replace_termcodes calls needs
to invoke replace_termcodes, and also check if RHS is equal to "<Nop>".
To reduce code duplication, factor this out into a helper function.
Also add an rhs_is_noop flag to MapArguments; buf_do_map_explicit
expects an empty {rhs} string for "<Nop>", but also needs to distinguish
that from something like ":map lhs<cr>" where no {rhs} was provided.
- getchar.c: Use allocated buffer for rhs in MapArgs
Since the MAXMAPLEN limit does not apply to the RHS of a mapping (or
else an RHS that calls a really long autoload function from a plugin
would be incorrectly rejected as being too long), use an allocated
buffer for RHS rather than a static buffer of length MAXMAPLEN + 1.
- Mappings LHS and RHS can contain literal space characters, newlines, etc.
- getchar.c: replace_termcodes in str_to_mapargs
It makes sense to do this; str_to_mapargs is, intuitively, supposed to
take a "raw" command string and parse it into a totally "do_map-ready"
struct.
- api/vim.c: Update lhs, rhs len after replace_termcodes
Fixes a bug in which replace_termcodes changes the length of lhs or rhs,
but the later search through the mappings/abbreviations hashtables
still uses the old length value. This would cause the search to fail
erroneously and throw 'E31: No such mapping' errors or 'E24: No such
abbreviation' errors.
- getchar: Create new map_arguments struct
So that a string of map arguments can be parsed into a more useful, more
portable data structure.
- getchar.c: Add buf_do_map function
Exactly the same as the old do_map, but replace the hardcoded references
to the global `buf_T* curbuf` with a function parameter so that we can
invoke it from nvim_buf_set_keymap.
- Remove gettext calls in do_map error handling
Before now, Nvim always degrades UI capabilities to the lowest-common
denominator. For example, if any connected UI has `ext_messages=false`
then `ext_messages=true` requested by any other connected UI is ignored.
Now `nvim_ui_attach()` supports `override=true`, which flips the
behavior: if any UI requests an `ext_*` UI capability then the
capability is enabled (and the legacy behavior is disabled).
Legacy UIs will be broken while a `override=true` UI is connected, but
it's useful for debugging: you can type into the TUI and observe the UI
events from another connected (UI) client. And the legacy UI will
"recover" after the `override=true` UI disconnects.
Example using pynvim:
>>> n.ui_attach(2048, 2048, rgb=True, override=True, ext_multigrid=True, ext_messages=True, ext_popupmenu=True)
>>> while True: n.next_message();
Problem: No simple way to label quickfix entries.
Solution: Add the "module" item, to be used instead of the file name for
display purposes. (Martin Szamotulski)
d76ce85266
On Windows we store non-config data in "$XDG_DATA_HOME/nvim-data". But
the "…/site" items in 'runtimepath' did not correctly point to that
location, they used "…/nvim/site".
Fix the init logic to use "…/nvim-data/site".
closes#9910
Callers can instead specify `args_rm={'--headless'}`.
TODO: should `nvim_argv` have "--headless" by default? Need to inspect
some uses of spawn(nvim_argv) ...
Problem: Using `:stopinsert` while in normal mode in a terminal buffer
prevents neovim from entering insert mode.
Solution: Move `stop_insert_mode = false` from terminal_check to
terminal_enter to be consistent with edit.c, as suggested by bfredl in
#9889.
Closes https://github.com/neovim/neovim/issues/9889.
- Allow floating windows of width 1. #9846
- For a new floating window the size must be specified. Later on we
might try to calculate a reasonable size by buffer contents
- Remember the configured size of a window, just like its position.
- Make get_config and set_config more consistent. Handle relative='' properly in set_config.
get_config doesn't return keys that don't make sense for a non-floating window.
- Don't use width=0 for non-changed width, just omit the key.
vim-patch:8.0.0714: when a timer causes a command line redraw " goes missing
Problem: When a timer causes a command line redraw the " that is displayed
for CTRL-R goes missing.
Solution: Remember an extra character to display.
a92522fbf3
vim-patch:8.0.0720: unfinished mapping not displayed when running timer
Problem: Unfinished mapping not displayed when running timer.
Solution: Also use the extra_char while waiting for a mapping and digraph.
(closesvim/vim#1844)
6a77d2667eclose#9835
Using `:wincmd j` and friends doesn't make much sense to a floating window. For
convenience though, any direction will simply change to the previous window.
Make sure the previous window is valid, not the current window, and not another
floating window. Change to the first window (which is never a floating window)
otherwise.
validate_cursor() is called regularly, but only for the current window.
When changing the buffer for a non-current window, we need to invoke it
in the context of that window.
CA_COMMAND_BUSY in nv_event() was carried over from Vim nv_cursorhold()
(ref: e5165bae11). It prevents :startinsert from working during a RPC
call, so remove it.
Helped-by: glacambre <me@r4>
closes#7254
Problem: Calling :stopinsert from RPC while in terminal-mode does not
go back to normal-mode.
Solution: Implement a check() handler for state_enter(), adapted from
insert_check().
Fix#7807
Problem: Extending sign and foldcolumn below the text is confusing.
Solution: Let the sign and foldcolumn stop at the last text line, just like
the line number column. Also stop the command line window leader.
(Christian Brabandt)
8ee4c01b8c
Closes https://github.com/neovim/neovim/issues/9613
- input: recognize <kEqual>, <kComma>
- terminal.c: If we need to support function key, a change must be made
in libvtermkey. Currently, it emulates strictly VT220 terminal, and
returning numeric value in 'normal' mode is the expected behaviour.
closes#9810
- K_KORIGIN instead of K_KCENTER: This name is similar to what is used
by xev. Alternative could be K_KBEGIN as hinted here:
https://invisible-island.net/xterm/ctlseqs/ctlseqs.html#h2-PC-Style-Function-Keys
But I find Begin and Home too similar, and it might induced some
confusion. The naming looked related to some old keyboard
configuration.
- keymap.c: alias KPPeriod to kDel instead of kPoint.
This might seems weird, but this is actually the behaviour that should
be expected. libtermkey produces "KPPeriod" when num lock is off. To
fix this would need to change this name in termkey.
closes#9780closes#9793
closes#990closes#9295
- Support for multiple auto-adjusted sign columns.
With this change, having more than one sign on a line, and with the
'auto' setting on 'signcolumn', extra columns will shown automatically
to accomodate all the existing signs.
For example, suppose we have this view:
5147 }
5148
5149 return sign->typenr;
5150 }
5151 }
5152 return 0;
5153 }
5154
We have GitGutter installed, so it tells us about modified lines that
are not commmited. So let's change line 5152:
5147 }
5148
5149 return sign->typenr;
5150 }
5151 }
~ 5152 return 0;
5153 }
5154
Now we add a mark over line 5152 using 'ma' in normal mode:
5147 }
5148
5149 return sign->typenr;
5150 }
5151 }
a ~ 5152 return 0;
5153 }
5154
Previously, Vim/Nvim would have picked only one of the signs,
because there was no support for having multiple signs in a line.
- Remove signs from deleted lines.
Suppose we have highlights on a group of lines and we delete them:
+ 6 use std::ops::Deref;
--+ 7 use std::borrow::Cow;
--+ 8 use std::io::{Cursor};
9 use proc_macro2::TokenStream;
10 use syn::export::ToTokens;
--+ 11 use std::io::Write;
>> 12 use std::ops::Deref;
Without this change, these signs will momentarily accumulate in
the sign column until the plugins wake up to refresh them.
+ --+ --+ --+ >> 6
Discussion: It may be better to extend the API a bit and allow this
to happen for only certain types of signs. For example, VIM marks
and vim-gitgutter removal signs may want to be presreved, unlike
line additions and linter highlights.
- 'signcolumn': support 'auto:NUM' and 'yes:NUM' settings
- sort signs according to id, from lowest to highest. If you have
git-gutter, vim-signature, and ALE, it would appear in this order:
git-gutter - vim-signature - ALE.
- recalculate size before screen update
- If no space for all signs, prefer the higher ids (while keeping the
rendering order from low to high).
- Prevent duplicate signs. Duplicate signs were invisible to the user,
before using our extended non-standard signcolumn settings.
- multi signcols: fix bug related to wrapped lines.
In wrapped lines, the wrapped parts of a line did not include the extra
columns if they existed. The result was a misdrawing of the wrapped
parts. Fix the issue by:
1. initializing the signcol counter to 0 when we are on a wrap boundary
2. allowing for the draw of spaces in that case.
Nvim doesn't expect a window-changing command on an created-window event.
autocmd WinNew * wincmd p
help help
- A snapshot for window 1000 is created.
- The window is split and the cursor changes to the new window 1001.
- The autocmd kicks in and switches back to 1000.
- The help buffer is opened.
- On closing the help window 1000, it tries to go back to the snapshotted window
which is... 1000.
- wp1000->w_buffer == NULL
- w_buffer is used by check_cursor()
- 🧨 -> 💥
Fixes https://github.com/neovim/neovim/issues/9773
- Lua test correctly fails when 8.1.0849 is reverted.
- 8.1.1001 bug does not manifest in Neovim.
vim-patch:8.1.0849: cursorline highlight is not always updated
Problem: Cursorline highlight is not always updated.
Solution: Set w_last_cursorline when redrawing. Fix resetting cursor flags
when using the popup menu.
c07ff5c60a
vim-patch:8.1.1001: Visual area not correct when using 'cursorline'
Problem: Visual area not correct when using 'cursorline'.
Solution: Update w_last_cursorline also in Visual mode. (Hirohito Higashi,
closesvim/vim#4086)
8156ed3755
bisected to f5d5da3917
Other test steps:
nvim -u NORC
:terminal tree / " Produces lots of output
:edit somefile.txt
:vsplit
:vsplit
<c-w>l
<c-w>l
<c-w>h
<c-w>p
Previous approach skipped the test if the expected value matched the
default value ("dark"). New approach always checks, but uses retry() to
ignore potentially wrong 'background' before the terminal response is
handled.
Problem: If autocmd pattern only contained `++once` handlers, and
all of them completed, then there would be an empty group
displayed by `:autocmd Foo`.
Solution: Delete the pattern if all of its commands were deleted.
If terminal response is received during startup, set 'background' from
a nested "one-shot" (once) VimEnter autocmd.
The previous not-so-clever "self-rescheduling" approach could cause
a long delay at startup (event-loop does not make forward progress).
fixes#9675
ref #9509
Adds a new feature to :autocmd which sets the handler to be executed at
most one times.
Before:
augroup FooGroup
autocmd!
autocmd FileType foo call Foo() | autocmd! FooGroup * <buffer>
augroup END
After:
autocmd FileType foo once call Foo()
Problem: Relative cursor position is not calculated correctly.
Solution: Always set topline, also when window is one line only.
(Robert Webb) Add more info to getwininfo() for testing.
8fcb60f961
Since uv_os_setenv uses SetEnvironmentVariableW, _wenviron is no
updated. As a result, inconsistency occurs in completion of environment
variable names. Change to use GetEnvironmentStaringsW instead of
_wenviron to solve it.
Note: the test fails on non-Windows CI (Travis linux, Quickbuild bsd):
even on master before the env.c changes in this patch-series.
Maybe the unix part of printenv-test.c needs to be revisited.
Signed-off-by: Justin M. Keyes <justinkz@gmail.com>
It's reported that the Windows widechar variants do automatically
convert from the current codepage to UTF16, which is very helpful. So
the "widechar" impls are a good direction. But libuv v1.12 does that
for us, so the next commit will use that instead.
ref #8398
ref #9267
- Like Vim, use set_option_value() followed by reset_option_was_set().
- Do not use set_string_default(), so the default is predictable.
This affects `:set bg&`.
- Wait until end-of-startup (VimEnter) to handle the response. The
response is racey anyways, so timing is irrelevant. This allows
OptionSet to be triggered, unlike during startup.
Loading existing files into a buffer is non-trivial and requires a window.
Creating an unnamed emtpy buffer is trivial and safe though, thus worth a
special case.
Change nvim_buf_set_option to use aucmd_prepbuf. This is necessary
to allow some options to be set on a not yet displayed buffer, such
as 'buftype' option.
vim-patch:7.4.1858: Add BLN_NEW to enforce buflist_new creating new buffer
Why?
- Because we can.
- Because the TUI is just another GUI™
- Because it looks kinda nice, and provides useful context like 1 out of 100
times
Complies with "don't pay for what you don't use".
Some crashes for resizing were unfolded, add tests for those.
Makes the 'scrollback' option more consistent (same default for all buffers) and future-proof.
- Default to -1 for all buffers, but treat it as an implementation detail.
- Document range of 1 - 100_000.
- New terminal buffer by default sets scrollback=10_000 if the global default is -1.
- Existing terminal buffer: On entering terminal-mode or on refresh, if the user explicitly did `:set[local] scbk=-1`, the local value goes to 100_000 (max). (This is undocumented on purpose. Users should work with explicit values in the range of 1-100_000.)
- Avoid using platform-specific shell, it failed in MINGW_64 env.
- tty-test.c echos our input, which is exactly what we need for this test.
- Test fails correctly if 894f6bee54 is reverted.
wp->w_height_inner now contains the "inner" size, regardless if the
window has been drawn yet or not. It should be used instead of
wp->w_grid.Rows, for stuff that is not directly related to accessing
the allocated grid memory, such like cursor movement and terminal size
- Return the menu properties, not only its children.
- If the {path} param is given, return only the first node. The "next"
nodes in the linked-list are irrelevant.
:menu should print sub-menu contents. E.g. this should print the
"File.Save" submenu:
nvim -u NORC
:source $VIMRUNTIME/menu.vim
:menu File.Save
Regressed in dc685387a3
Blocks #8173
menu_get() also was missing some results for some cases.
Previously the mouse tests set 'listchars', but not 'list'. Funnily enough, the
space, where the `$` would normally appear, would still use new highlight group.
Set 'list' for good and fix the tests accordingly.
In vim, scrolling a window might mess up the cmdline. To keep it simple,
cmdline was always cleared for any window scroll. In nvim, where safe scrolling
is implemented in the TUI layer, this problem doesn't exist.
Clearing the message on scrolling, when we not do it e.g when switching tabs
is a bit weird, as the former is a much smaller context change.
A vim patch introduced the possibility to avoid the cmdlline clear for
redraws caused by async events. This case will now trivially be covered,
as the redraw is always avoided.
vim-patch:8.0.0592: if a job writes to a buffer screen is not updated
Implement lazy loading for vim.submodule, this would be over-engineering
for inspect only, but we expect to use this solution also for more and
larger modules.
