In https://github.com/neovim/neovim/pull/22214, init_default_autocmds
has been turned into a lua function call to nvim_create_augroup and
nvim_create_autocmd.
This introduced a strange regression: a test in vim_spec.lua started
failing with its last_set_chan value switching from 0 to
-9223372036854775808.
It turns out that -9223372036854775808 is the value of LUA_INTERNAL_CALL
and would be inherited as last_set_chan by options set from the command
line due to the WITH_SCRIPT_CONTEXT macro not restoring the channel id
(WITH_SCRIPT_CONTEXT is used by nvim_create_augroup).
While the new RPC encoder/decoder implementation in general should have
less overhead than the deleted UIBridge serializer previously used for
the TUI, it regresses on redraw latency in one important aspect.
The old bridge implementation allowed the TUI to process a
previous screen line internally in parallel with the main thread
rendering the next one in win_line etc. As printing the escape
sequences in highlighted cells has a considerable hit in profiles,
this has a substantial effect on redraw latency. The RPC implementation,
however, waits with sending any data until either a flush, or the buffer is full.
This change lowers the granularity of communication again, using an
adjustable threshold counted in number of cell events (discounting
long repeats and clearing as maximum a single extra event).
The current value is guesstimated to something simple on a reasonable
scale, which should be bigger than a single line, but multiple events
for a big multi-window screen.
Problem:
The "force" flag of win_close() complicates the code and adds edge cases
where it is not clear what the correct behavior should be.
The "free_buf" flag of win_close() is passed on to float windows when
closing the last window of a tabpage, which doesn't make much sense.
Solution:
Remove the "force" flag and always close float windows as if :close! is
used when closing the last window of a tabpage, and set the "free_buf"
flag for a float window based on whether its buffer can be freed.
As 'hidden' is on by default, this change shouldn't affect many people.
Problem:
Build is not reproducible, because generated source files (.c/.h/) are not
deterministic, mostly because Lua pairs() is unordered by design (for security).
https://github.com/LuaJIT/LuaJIT/issues/626#issuecomment-707005671https://www.lua.org/manual/5.1/manual.html#pdf-next
> The order in which the indices are enumerated is not specified [...]
>
>> The hardening of the VM deliberately randomizes string hashes. This in
>> turn randomizes the iteration order of tables with string keys.
Solution:
- Update the code generation scripts to be deterministic.
- That is only a partial solution: the exported function
(funcs_metadata.generated.h) and ui event
(ui_events_metadata.generated.h) metadata have some mpack'ed
tables, which are not serialized deterministically.
- As a workaround, introduce `PRG_GEN_LUA` cmake setting, so you can
inject a modified build of luajit (with LUAJIT_SECURITY_PRN=0)
that preserves table order.
- Longer-term we should change the mpack'ed data structure so it no
longer uses tables keyed by strings.
Closes#20124
Co-Authored-By: dundargoc <gocdundar@gmail.com>
Co-Authored-By: Arnout Engelen <arnout@bzzt.net>
Problem:
No easy way to position a LSP hover window relative to mouse.
Solution:
Introduce another option to the `relative` key in `nvim_open_win()`.
With this PR it should be possible to override the handler and do something
similar to this https://github.com/neovim/neovim/pull/19481#issuecomment-1193248674
to have hover information displayed from the mouse.
Test case:
```lua
local util = require('vim.lsp.util')
local function make_position_param(window, offset_encoding)
window = window or 0
local buf = vim.api.nvim_win_get_buf(window)
local row, col
local mouse = vim.fn.getmousepos()
row = mouse.line
col = mouse.column
offset_encoding = offset_encoding or util._get_offset_encoding(buf)
row = row - 1
local line = vim.api.nvim_buf_get_lines(buf, row, row + 1, true)[1]
if not line then
return { line = 0, character = 0 }
end
if #line < col then
return { line = 0, character = 0 }
end
col = util._str_utfindex_enc(line, col, offset_encoding)
return { line = row, character = col }
end
local make_params = function(window, offset_encoding)
window = window or 0
local buf = vim.api.nvim_win_get_buf(window)
offset_encoding = offset_encoding or util._get_offset_encoding(buf)
return {
textDocument = util.make_text_document_params(buf),
position = make_position_param(window, offset_encoding),
}
end
local hover_timer = nil
vim.o.mousemoveevent = true
vim.keymap.set({ '', 'i' }, '<MouseMove>', function()
if hover_timer then
hover_timer:close()
end
hover_timer = vim.defer_fn(function()
hover_timer = nil
local params = make_params()
vim.lsp.buf_request(
0,
'textDocument/hover',
params,
vim.lsp.with(vim.lsp.handlers.hover, {
silent = true,
focusable = false,
relative = 'mouse',
})
)
end, 500)
return '<MouseMove>'
end, { expr = true })
```
Problem: Unable to customize the column next to a window ('gutter').
Solution: Add 'statuscolumn' option that follows the 'statusline' syntax,
allowing to customize the status column. Also supporting the %@
click execute function label. Adds new items @C and @s which
will print the fold and sign columns. Line numbers and signs
can be clicked, highlighted, aligned, transformed, margined etc.
- The defined interface for the UI is only the RPC protocol. The original
UI interface as an array of function pointers fill no function.
- On the server, all the UI:s are all RPC channels.
- ui.c is only used on the server.
- The compositor is a preprocessing step for single-grid UI:s
- on the client, ui_client and tui talk directly to each other
- we still do module separation, as ui_client.c could form the basis
of a libnvim client module later.
Items for later PR:s
- vim.ui_attach is still an unhappy child, reconsider based on plugin experience.
- the flags in ui_events.in.h are still a mess. Can be simplified now.
- UX for remote attachment needs more work.
- startup for client can be simplified further (think of the millisecs we can save)
This is needed for #18375 for the obvious reasons.
note: verbose_terminfo_event is only temporarily needed
until the full TUI process refactor is merged.
This is cherry-picked from these Vim patches:
Only applicable change outside vi_diff.txt in patch 8.1.1226:
6c60f47fb9
Most changes outside starting.txt and vi_diff.txt in patch 8.1.1280:
25c9c680ec
Missing docs for 'mousemoveevent':
cbaff5e06e
* fix: log and clear error in ui_comp_event
* fix: handling error in each map_foreach_value iteration
* fix: handling error decl in for_each loop
* fix: updating initerr to const, removing initerr free-ing
* fix: using ERROR_SET for error check
* fix: wrapping ERROR_INIT in parens to allow for including inside macro
Problem: Some source files are too big.
Solution: Move buffer and window related functions to evalbuffer.c and
evalwindow.c. (Yegappan Lakshmanan, closesvim/vim#4898)
261f346f81
* fix(PVS/V009): start file with special comment
* fix(PVS/V501): identical sub-expressions for comparison
* fix(PVS/V560): part of conditional expression is always true/false
* fix(PVS/V593): review expression of type A = B < C
* fix(PVS/V614): potentially uninitialized variable used
Allow Include What You Use to remove unnecessary includes and only
include what is necessary. This helps with reducing compilation times
and makes it easier to visualise which dependencies are actually
required.
Work on https://github.com/neovim/neovim/issues/549, but doesn't close
it since this only works fully for .c files and not headers.
Problem: Handling 'statusline' errors is spread out.
Solution: Pass the option name to the lower levels so the option can be
reset there when an error is encountered. (Luuk van Baal,
closesvim/vim#11467)
7b224fdf4a
