XXX: comment at https://stackoverflow.com/q/1173342 :
> Windows recycles PIDs quite fast, you have to be extra careful not
> to kill unrelated processes. These APIs will report PPIDs for long
> dead processes whose PIDs may have been recycled. Check the parent
> start date to make sure it is related to the processes you spawned.
UV_PROCESS_DETACHED compels libuv:uv__process_child_init() to call
setsid() in the child just after fork(). That ensures the process and
its descendants are grouped in a separate session (and process group).
The following jobstart() call correctly groups `sh` and `sleep` in a new
session (and process-group), where `sh` is the "session leader" (and
process-group leader):
:call jobstart(['sh','-c','sleep 60'])
SESN PGRP PID PPID Command
30383 30383 30383 3620 │ ├─ -bash
30383 31432 31432 30383 │ │ └─ nvim -u NORC
30383 31432 31433 30383 │ │ ├─ nvim -u NORC
8105 8105 8105 31432 │ │ └─ sh -c sleep 60
8105 8105 8106 8105 │ │ └─ sleep 60
closes#6530
ref: https://stackoverflow.com/q/1046933
ref: https://unix.stackexchange.com/a/404065
Helped-by: Marco Hinz <mh.codebro+github@gmail.com>
Discussion
------------------------------------------------------------------------
On my linux box before this patch, the termclose_spec.lua:'kills job
trapping SIGTERM' test indirectly causes cmake/busted to wait for 60s.
That's because the test spawns a `sleep 60` descendant process which
hangs around even after nvim exits: nvim killed the parent PID, but not
PGID (process-group), so the grandchild "reparented" to init (PID 1).
Session contains processes (and process-groups) which are logically part
of the same "login session". Process-group is a set of
logically/informally-related processes within a session; for example,
shells assign a process group to each "job". Session IDs and PGIDs both
have type pid_t (like PIDs).
These OS-level mechanisms are, as usual, legacy accidents whose purpose
is upheld by convention and folklore. We can use session-level grouping
(setsid), or we could use process-group-level grouping (setpgid).
Vim uses setsid() if available, otherwise setpgid(0,0).
Windows
------------------------------------------------------------------------
UV_PROCESS_DETACHED on win32 sets CREATE_NEW_PROCESS_GROUP flag.
But uv_kill() does not kill the process-group:
https://github.com/nodejs/node/issues/3617
Ideas:
- Set UV_PROCESS_DETACHED (CREATE_NEW_PROCESS_GROUP), then call
GenerateConsoleCtrlEvent(CTRL_BREAK_EVENT, pid)
- Maybe won't work because MSDN says "Only processes that share the
same console as the calling process receive the signal."
https://docs.microsoft.com/en-us/windows/console/generateconsolectrlevent
But CREATE_NEW_PROCESS_GROUP creates a new console ...
ref https://stackoverflow.com/q/1453520
- Group processes within a "job". libuv does that *globally* for
non-detached processes: uv__init_global_job_handle.
- Iterate through CreateToolhelp32Snapshot().
- https://stackoverflow.com/q/1173342
- Vim does this, see terminate_all()
This means that the major way that :lmap mappings are applied works as
one would expect with macros.
This also means that having a translation with 'keymap' does not
preclude using mappings in insert mode with :imap.
If the mental model of :lmap mappings is a translation between your
keyboard and vim proper, then they should take preference over :imap
(and other) mappings. This patch makes that happen.
The mental model of :lmap and 'keymap' is of a transformation done
before anything else. Hence when recording a macro, or writing to a
scriptfile, the transformed keys should be recorded instead of the keys
before the transformation.
vim-patch:8.0.0262: Farsi support is barely tested
Problem: Farsi support is barely tested.
Solution: Add more tests for Farsi. Clean up the code.
ddf662a1c8
vim-patch:8.0.0263: Farsi support is not tested enough
Problem: Farsi support is not tested enough.
Solution: Add more tests for Farsi. Clean up the code.
80627cf51f
Problem: Getting name of cleared highlight group is wrong. (Matt Wozniski)
Solution: Only skip over cleared names for completion. (closesvim/vim#1592)
Also fix that a cleared group causes duplicate completions.
c96272e30e
Problem: Completion for :match does not show "none" and other missing
highlight names.
Solution: Skip over cleared entries before checking the index to be at the
end.
15eedf1d62
vim-patch:8.0.0234: crash when using put in Visual mode
Problem: When several lines are visually selected and one of them is short,
using put may cause a crash. (Axel Bender)
Solution: Check for a short line. (Christian Brabandt)
941c12da3c
vim-patch:8.0.0236: gcc complains about uninitialized variable
Problem: Gcc complains that a variable may be used uninitialized. Confusion
between variable and label name. (John Marriott)
Solution: Initialize it. Rename end to end_lnum.
6a717f17ec
vim-patch:8.0.0225: put in Visual block mode terminates early
Problem: When a block is visually selected and put is used on the end of
the selection only one line is changed.
Solution: Check for the end properly. (Christian Brabandt, neovim issue
5781)
9957a10d0f
With MSVC, STDOUT_FILENO and STDERR_FILENO are defined as function calls instead of constants, meaning they can't be assigned to enum values. The enum was only used in one file, so it has been removed. A definition for STDIN_FILENO has been added that is consistent with the other two definitions.
The following code..
au VimEnter,DirChanged * if filereadable('.git/cscope.out') |
\ exe 'cs add .git/cscope.out' | endif
..would lead to this issue:
Error detected while processing VimEnter Auto commands for "*":
cs_read_prompt EOF: Interrupted system call
Error detected while processing VimEnter Auto commands for "*":
E262: error reading cscope connection 0
A signal, in this case SIGCHLD, during a system call leads to errno being set
to EINTR. Ignore it.
This is merely a workaround for the time being. We don't block SIGCHLD signals,
since they're needed by libuv. The proper fix would be to rewrite if_cscope.c to
use libuv for handling processes.
Problem: If cscope fails a search Vim may hang.
Solution: Bail out when a search error is encountered. (Safouane Baroudi,
closesvim/vim#2598)
1274d33493
Currently writedelay shows the sequence of characters that are sent to
the UI/TUI module. Here nvim has already applied an optimization: when
attempting to put a char in a screen cell, if the same char already was
there with the same attributes, UI output is disabled. When debugging
redrawing it it sometimes more useful to inspect the redraw stream one
step earlier, what region of the screen nvim actually is recomputing
from buffer contents (win_line) and from evaluating statusline
expressions.
Take the popupmenu as an example. When closing the popupmenu (in the
TUI), currently 'writedelay' looks like vim only is redrawing the region
which the pum covered. This is not what happens internally: vim redraws
the entire screen, even if only outputs the changed region.
This commit allows negative values of 'writedelay', which causes a delay
for all redrawn characters, even if the character already was displayed
by the UI before.
Most fonts should have these by now. Both are a significant visual
improvement.
- Vertical connecting bar `│` is used by tmux, pstree, Windows 7 cmd.exe
and nvim-qt.exe.
- Middle dot `·` works on Windows 7 cmd.exe, nvim-qt.exe.
For reference: tmux uses these chars to draw lines: │ ├ ─
Problem: When a block is visually selected and put is used on the end of
the selection only one line is changed.
Solution: Check for the end properly. (Christian Brabandt, neovim issue
5781)
9957a10d0f
Note: code change was covered by c2a1821611
Problem: When a multi-byte character ends in a zero byte, putting blockwise
text puts it before the character instead of after it.
Solution: Use int instead of char for the character under the cursor.
(Luchr, closesvim/vim#1403) Add a test.
c81299684b
vim-patch:8.0.0214: leaking memory when syntax cluster id is unknown
Problem: Leaking memory when syntax cluster id is unknown. (Coverity)
Solution: Free the memory.
d7a96151e0
Problem: The buffer used to store a key name theoreticaly could be too
small. (Coverity)
Solution: Count all possible modifier characters. Add a check for the
length just in case.
423977d3ce
