fs_init() must be called before early_init() in init/helpers.lua
If I run 'make unittest' on my Mac (macOS 10.14/Mojave or 12/Big Sur, intel
CPU), every test produce a core dump.
Call sequence in the core is:
early_init() main.c:197
set_init_1() option.c:508
runtimepath_default() runtime.c:1205
get_lib_dir() runtime.c:1175
os_isdir() fs.c:137
os_getperm() fs.c:777
os_stat() fs.c:761
fs_loop_lock() fs.c:72
uv_mutex_lock(&fs_loop_mutex) thread.c:352
abort()
.deps/build/src/libuv/src/unix/thread.c:
void uv_mutex_lock(uv_mutex_t* mutex) {
if (pthread_mutex_lock(mutex))
abort(); // line 352
}
So pthread_mutex_lock(&fs_loop_mutex) failed. The reason seems to be simple.
fs_init() was not called and fs_loop_mutex has not been initialized. fs_init()
was moved out from early_init() in main.c by
b87867e69e, but unit/helpers.lua was not updated
accordingly.
Most code in keymap.h is for keycode definitions, while most code in
keymap.c is for the parsing and conversion of keycodes.
The name "keymap" may also make people think these two files are for
mappings, while in fact keycodes are used even when no mappings are
involved, so "keycodes" should be a better file name than "keymap".
Problem: GUI tests fail because the test doesn't use a modifier.
Solution: Add "\{xxx}" to be able to encode a modifier.
ebe9d34aa0
Change macros to enums to use them in unit tests.
Problem: Cannot map <C-H> when modifyOtherKeys is enabled.
Solution: Add the <C-H> mapping twice, both with modifier and as 0x08. Use
only the first one when modifyOtherKeys has been detected.
459fd785e4
Add REPTERM_NO_SPECIAL instead of REPTERM_SPECIAL because the meaning of
"special" is different between Vim and Nvim.
Omit seenModifyOtherKeys as Nvim supports attaching multiple UIs.
Omit tests as they send terminal codes.
Keep the behavior of API functions.
Ref: #9342
Adds the option to have a single global statusline for the current window at the bottom of the screen instead of a statusline at the bottom of every window. Enabled by setting `laststatus = 3`.
Due to the fact that statuslines at the bottom of windows are removed when global statusline is enabled, horizontal separators are used instead to separate horizontal splits. The horizontal separator character is configurable through the`horiz` item in `'fillchars'`. Separator connector characters are also used to connect the horizontal and vertical separators together, which are also configurable through the `horizup`, `horizdown`, `vertleft`, `vertright` and `verthoriz` items in `fillchars`.
The window separators are highlighted using the `WinSeparator` highlight group, which supersedes `VertSplit` and is linked to `VertSplit` by default in order to maintain backwards compatibility.
This includes a partial port of Vim patch 8.2.2569 and some changes to
nvim_eval_statusline() to allow a multibyte fillchar. Literally every
line of C code touched by that patch has been refactored in Nvim, and
that patch contains some irrelevant foldcolumn tests I'm not sure how to
port (as Nvim's foldcolumn behavior has diverged from Vim's).
marktree.c was originally constructed as a "generic" datatype,
to make the prototyping of its internal logic as simple as possible
and also as the usecases for various kinds of extmarks/decorations was not yet decided.
As a consequence of this, various extra indirections and allocations was
needed to use marktree to implement extmarks (ns/id pairs) and
decorations of different kinds (some which is just a single highlight
id, other an allocated list of virtual text/lines)
This change removes a lot of indirection, by making Marktree specialized
for the usecase. In particular, the namespace id and mark id is stored
directly, instead of the 64-bit global id particular to the Marktree
struct. This removes the two maps needed to convert between global and
per-ns ids.
Also, "small" decorations are stored inline, i.e. those who
doesn't refer to external heap memory anyway. That is highlights (with
priority+flags) are stored inline, while virtual text, which anyway
occurs a lot of heap allocations, do not. (previously a hack was used
to elide heap allocations for highlights with standard prio+flags)
TODO(bfredl): the functionaltest-lua CI version of gcc is having
severe issues with uint16_t bitfields, so splitting up compound
assignments and redundant casts are needed. Clean this up once we switch
to a working compiler version.
Co-authored-by: Sean Dewar <seandewar@users.noreply.github.com>
Co-authored-by: Gregory Anders <greg@gpanders.com>
Co-authored-by: Sebastian Volland <seb@baunz.net>
Co-authored-by: Lewis Russell <lewis6991@gmail.com>
Co-authored-by: zeertzjq <zeertzjq@outlook.com>
Problem: Problem with :cd when editing file in non-existent directory. (Yee
Cheng Chin)
Solution: Prepend the current directory to get the full path. (closesvim/vim#8903)
c6376c7984
Problem: Items in a list given to :const can still be modified.
Solution: Work like ":lockvar! name" but don't lock referenced items.
Make locking a blob work.
021bda5671
Problem: Cannot use octal numbers in scriptversion 4.
Solution: Add the "0o" notation. (Ken Takata, closesvim/vim#5304)
c17e66c5c0
:scriptversion is N/A.
Cherry-pick latest str2nr() doc changes from v8.1.2035.
Cherry-pick various mentions of the 0o prefix from:
- v8.2.2324
- 2346a63784
- 11e3c5ba82
- 82be4849ee
Patch used ascii_isbdigit() by mistake, which was fixed in v8.2.2309.
Make STR2NR_OOCT work the same as STR2NR_OCT when forcing.
In Vim, STR2NR_FORCE | STR2NR_OOCT isn't handled, and doesn't actually
force anything. Rather than abort(), make it work as STR2NR_OCT.
This means STR2NR_FORCE | STR2NR_OCT works the same as
STR2NR_FORCE | STR2NR_OOCT and STR2NR_FORCE | STR2NR_OCT | STR2NR_OOCT.
Problem: The str2nr() tests fail.
Solution: Add missing part of patch.
1ac90b4fa6
Add extra tests for quoted numbers in vim_str2nr_spec.lua, as the
included ones in this patch are somewhat lacking.
Problem: Obvious mistakes are accepted as valid expressions.
Solution: Be more strict about parsing numbers. (Yasuhiro Matsumoto,
closesvim/vim#3981)
16e9b85113
Update vim_str2nr_spec.lua to add more tests that use strict = true.
Remove syncolor.vim in favor of defining the default highlight groups
directly in `init_highlight`. This approach provides a number of
advantages:
1. The highlights are always defined, regardless of whether or not the
syntax regex engine is enabled.
2. Redundant sourcing of syntax files is eliminated (syncolor.vim was
often sourced multiple times based on how the user's colorscheme file
was written).
3. The syntax highlighting regex engine and the highlight groups
themselves are more fully decoupled.
4. Removal of the confusing `:syntax on` / `:syntax enable` dichotomy
(they now both do the same thing).
This approach also correctly solves a number of bugs related to
highlighting (#15176, #12573, #15205).
As per #14236, performing extmark cleanup in a certain namespace does
not guarantee removing all the extmarks inside given namespace.
The issue resides within the tree node removal method and results in
a couple of rare edge cases.
To demonstrate what causes this bug, I'll give an example covering one
of the edge cases.
=== AN EXAMPLE ===
(A) (B) (C) (D) (E)
--------- --------- --------- --------- ---------
<0, 1> <0, 1> <0, 1> <0, 1> <0, 1>
<0, 2> <0, 2> <0, 2> <0, 2> <0, 2>
<0, 3> <0, 3> <0, 3> <0, 3> <0, 3>
<0, 4> <0, 4> <0, 4> <0, 4> <0, 4>
<0, 5> <0, 5> <0, 5> <0, 5> <0, 5>
<0, 6> <0, 6> <0, 6> <0, 6> <0, 6>
<0, 7> <0, 7> <0, 7> <0, 7> <0, 7>
<0, 8> <0, 8> <0, 8> <0, 8> <0, 8>
<0, 9> <0, 9> * * <0, 9> * <0, 9>
[0, 10] * [0, 10] <0, 9> [0, 11] [0, 11]
[0, 11] [0, 11] [0, 11] [0, 12] [0, 12] *
[0, 12] [0, 12] [0, 12] [0, 13] [0, 13]
[0, 13] [0, 13] [0, 13] [0, 14] [0, 14]
[0, 14] [0, 14] [0, 14] [0, 15] [0, 15]
[0, 15] [0, 15] [0, 15] [0, 16] [0, 16]
[0, 16] [0, 16] [0, 16] [0, 17] [0, 17]
[0, 17] [0, 17] [0, 17] [0, 18] [0, 18]
[0, 18] [0, 18] [0, 18] [0, 19] [0, 19]
[0, 19] [0, 19] [0, 19] [0, 20] [0, 20]
[0, 20] [0, 20] [0, 20]
DIAGRAM EXPLANATION
* Every column is a state of the marktree at a certain stage.
* To make it simple, I don't draw the whole tree. What you see are
2 leftmost parent nodes ([0, 10], [0, 20]) and their children placed
in order `MarkTreeIter` would iterate through. From top to bottom.
* Numbers on this diagram represent extmark coordinates. Relative
positioning and actual mark IDs used by the marktree are avoided
for simplicity.
* 2 types of brackets around coordinates represent 2 different
extmark namespaces (`ns_id`s).
* '*' shows iterator position.
ACTUAL EXPLANATION
Let's assume, we have two sets of extmarks from 2 different plugins:
* Plugin1: <0, 1-9>
* Plugin2: [0, 10-20]
1. Plugin2 calls
`vim.api.nvim_buf_clear_namespace(buf_handle, ns_id, 0, -1)`
to clear all its extmarks which results in `extmark_clear` call.
2. The iteration process goes on ignoring extmarks with irrelevant
`ns_id` from Plugin1, until it reaches [0, 10], entering state (A).
3. At the end of cleaning up process, `marktree_del_itr` gets called.
This function is supposed to remove given node and, if necessary,
restructure the tree. Also, move the iterator to the next node.
The bug occurs in this function.
4. The iterator goes backwards to the node's last child, to put it
in the place of its deleted parent later. (B)
5. The parent node is deleted and replaced with its child node. (C)
6. Since now this node has 8 children, which is less than
`MT_BRANCH_FACTOR - 1`, it get's merged with the next node. (D)
7. Finally, since at (B) the iterator went backward, it goes forward
twice, skipping [0, 11] node, causing this extmark to persist,
causing the bug. (E)
ANALYSIS AND SOLUTION
The algorithm works perfectly when the parent node gets replaced by
its child, but no merging occurs. I.e. the exact same diagram,
but without the (D) stage. If not for (D), it would iterate to <0, 9>
and then to [0, 11]. So, iterating twice makes sense. The actual problem
is in (C) stage, because the iterator index isn't adjusted and still
pointing to no longer existent node. So my solution is to adjust
iterator index after removing the child node.
More info: https://github.com/neovim/neovim/pull/14719
Problem: Number of status line items is limited to 80.
Solution: Dynamically allocate the arrays. (Rom Grk, closesvim/vim#7181)
8133cc6bf4
The members of stl_item_T have not been prefixed with stl_ contrary to
the vim patch because the amount of stl_ prefixes on single lines of
code in that region was hurtful to readability.
Seems like redundant env var separators (";" on Windows) in $PATH can
cause weird behavior. From #7377:
> After some time, system(['win32yank', '-o']) and system('win32yank -o')
> start returning different results: specifically first returns an
> empty string.
>
> 1. $PATH weirdly contains double semicolon followed by path to the
> “installation directory” (unpacked directory from archive).
> 2. If I run `let $PATH=substitute($PATH, ';;', ';', 'g')` the problem is fixed.
close#7377
ref 224f99b85d
Problem: Comparing two NULL list fails.
Solution: Change the order of comparing two lists.
7b293c730b
N/A patches for version.c:
vim-patch:8.2.1187: terminal2 test sometimes hangs in the GUI on Travis
Problem: Terminal2 test sometimes hangs in the GUI on Travis.
Solution: Disable Test_zz2_terminal_guioptions_bang() for now.
c85156bb89
vim-patch:8.2.1188: memory leak with invalid json input
Problem: Memory leak with invalid json input.
Solution: Free all keys at the end. (Dominique Pellé, closesvim/vim#6443,
closesvim/vim#6442)
6d3a7213f5
vim-patch:8.2.1196: build failure with normal features
Problem: Build failure with normal features.
Solution: Add #ifdef.
83e7450053
vim-patch:8.2.1198: terminal2 test sometimes hangs in the GUI on Travis
Problem: Terminal2 test sometimes hangs in the GUI on Travis.
Solution: Move test function to terminal3 to see if the problem moves too.
a4b442614c
This is inspired by Atom's "marker index" data structure to efficiently
adjust marks to text insertions deletions, but uses a wide B-tree
(derived from kbtree) to keep the nesting level down.
Problem: Dict and list could be GC'ed while displaying error in a timer.
(Yasuhiro Matsumoto)
Solution: Block garbage collection when executing a timer. Add
test_garbagecollect_soon(). Add "no_wait_return" to
test_override(). (closesvim/vim#4571)
adc6714aac
Problem: Cannot recover from a swap file.
Solution: Do not expand environment variables in the swap file name.
Do not check the extension when we already know a file is a swap
file. (Ken Takata, closes 4415, closesvim/vim#4369)
99499b1c05
`tui_terminal_after_startup` gets called right after resuming from
suspending (via `Ctrl-z`) already (not delayed as with the startup
itself), and would set `waiting_for_bg_response` to false then directly.
This results in the terminal response not being processed then anymore,
and leaking into Neovim itself.
This changes it to try 5 times always, which means that it typically
would stop after a few characters of input from the user typically, e.g.
with tmux, which does not send a reply.
While it might be better to have something based on the time (e.g. only
wait for max 1s), this appears to be easier to do.
Fixes regression in 8a4ae3d.
* handle_background_color: short-circuit if handled already
* Unit tests for handle_background_color
* set waiting_for_bg_response to false in tui_terminal_after_startup
By then it should have been received.
