Making an environment variable empty can be a way of unsetting it for
platforms that don't support unsetenv(). In most cases, we treat empty
variables as having been unset. For all others, use os_env_exists().
- Create a private libuv loop instead of re-using uv_default_loop(), to
avoid conflict[1] with existing watcher(s) on the fd.
- Expose the global "input" fd as a getter instead of a mutable global.
[1] .deps/build/src/libuv/src/unix/core.c:833:
uv__io_stop: Assertion `loop->watchers[w->fd] == w' failed.
If stdin is non-blocking, many tools (e.g. cat(1), read(1)) which assume
that stdin is blocking, will break in odd ways:
read: read error: 0: Resource temporarily unavailable
cat: -: Resource temporarily unavailable
rm: error closing file
libuv puts stdin in nonblocking mode, and leaves it that way at exit
(this is apparently by design). So, before this commit, this always
works (because the shell clobbers O_NONBLOCK):
$ nvim --cmd q
$ read
...but these forms do _not_ work:
$ nvim --cmd q && read
$ echo foo | nvim --cmd q && read
$ nvim && read
After this commit, all of the above forms work.
Background:
437b4397b9 (diff-41f4d294430cd8c36538999d62681ae2)https://github.com/fish-shell/fish-shell/issues/176#issuecomment-15800155
- bash (and other shells: zsh, tcsh, fish), upon returning to the
foreground, always sets fd 0 back to blocking mode. This practice only
applies to stdin, _not_ stdout or stderr (in practice these fds may be
affected anyways).
- bash/zsh/tcsh/fish do _not_ restore the non-blocking status of stdin
when _resuming a job_.
- We do _not_ save/restore the original flags visible to
fcntl(F_[SG]ETFL), because (counterintuitively) that isn't expected.
Helped-by: oni-link <knil.ino@gmail.com>
Closes#2086Closes#2377
---
Note: The following implementation of stream_set_blocking() was
discarded, because it resulted in a failed libuv assertion[1]:
int stream_set_blocking(int fd, bool blocking)
{
uv_pipe_t stream;
uv_pipe_init(uv_default_loop(), &stream, 0);
uv_pipe_open(&stream, fd);
int retval = uv_stream_set_blocking((uv_stream_t *)&stream, blocking);
uv_close((uv_handle_t *)&stream, NULL);
return retval;
}
[1] .deps/build/src/libuv/src/unix/core.c:833: uv__io_stop: Assertion `loop->watchers[w->fd] == w' failed.
For any of these functions, if {cmd} is a string, execute
"&shell &shellcmdflag '{cmd}'", or simply {cmd} if it's a list.
In termopen(), if the 'name' option is not supplied, try to guess using
'{cmd}' (string) or {cmd}[0] (list). Simplify ex_terminal to use the
string form of termopen().
termopen: get name from argument
Convert list_to_argv to tv_to_argv.
Helped-by: Björn Linse <@bfredl>
Helped-by: oni-link <knil.ino@gmail.com>
Helped-by: Thiago de Arruda <@tarruda>
Without the casts*, the compiler rightly warns about the os_getenv
losing the qualifier. This refactor adds a variable to manage this
properly, and renames the original variables to increase clarity.
- In UNIX systems where unsetenv() is available, it is used. Otherwise
the variables are set with the empty string.
- New check HAVE_UNSETENV for unsetenv()
- Added unit test to env_spec.lua
We already use wrappers for allocation, the new `xfree` function is the
equivalent for deallocation and provides a way to fully replace the malloc
implementation used by Neovim.
Klib pools were used to improve allocation efficiency for some small objects,
but it is not a thread-safe approach. Thread safety in allocations will be
required for implementing #2371).
- Properly save job event deferring state for recursive calls
- Disable breakcheck while running. Breakcheck can invoke job callbacks
in unexpected places.
Reported by @fourjay, a codepath that causes event_poll() to run before
event_init() will trigger a segfault as the events list will not have
been initialized. Exiting immediately from event_init() causes nvim to
hang, so just exit before running the events.
fixes#2339
event_poll() always leaves the libuv I/O loop after one iteration. The
duration for how long the loop polls for I/O is given by the shortest
timeout of all active timers. Is no timer active, I/O is/could be polled
indefinitely.
To make sure our timer is still active when I/O polling begins,
a prepare handle is used to start the timer right before the polling.
But by only using a timer that restarts after its timeout was reached,
we also can assure that polling is done with (nearly) the same finite
timeout.
For a short explanation how the I/O loop is working, see
http://docs.libuv.org/en/latest/design.html#the-i-o-loop.
channel_write() uses a ref-counted buffer for writing. This buffer
should be released if it was used in "refcount" channel_write() calls.
But calling channel_write() on a closed channel would return early and
not decrease the refcount of the used buffer.
This also removes the `#elseif defined(MSWIN)` clause. Due to the
enclosing `if` block, we will never get to this point when src starts with
a '%', making the whole #elseif block dead code.
- Remove JobActivity autocmd and v:job_data variable
- Simplify `jobstart` to receive:
- An argument vector
- An optional dictionary which may contain any of the current `jobstart`
options plus `on_stdout`, `on_stderr` and `on_exit` callbacks.
- Refactor and add more job tests
- Update documentation
Instead of selecting stderr on startup if stdin is not a tty, first try reading
from it and only switch to stderr when reading fails. With this behavior we
support commands like:
```
echo q | nvim -es
```
and
```
ls *.md | xargs nvim
```
Fixed small bugs in rstream.c to make this happen.
Since all reads are queued by the event loop, we must also queue the exit event,
or else the process_close function can close the job streams before received
data is processed.
This commit integrates libvterm with Neovim and implements a terminal emulator
with nvim buffers as the display mechanism. Terminal buffers can be created
using any of the following methods:
- Opening a file with name following the "term://[${cwd}//[${pid}:]]${cmd}"
URI pattern where:
- cwd is the working directory of the process
- pid is the process id. This is just for use in session files where a pid
would have been assigned to the saved buffer title.
- cmd is the command to run
- Invoking the `:terminal` ex command
- Invoking the `termopen` function which returns a job id for automating the
terminal window.
Some extra changes were also implemented to adapt with terminal buffers. Here's
an overview:
- The `main` function now sets a BufReadCmd autocmd to intercept the term:// URI
and spawn the terminal buffer instead of reading the file.
- terminal buffers behave as if the following local buffer options were set:
- `nomodifiable`
- `swapfile`
- `undolevels=-1`
- `bufhidden=hide`
- All commands that delete buffers(`:bun`, `:bd` and `:bw`) behave the same for
terminal buffers, but only work when bang is passed(eg: `:bwipeout!`)
- A new "terminal" mode was added. A consequence is that a new set of mapping
commands were implemented with the "t" prefix(tmap, tunmap, tnoremap...)
- The `edit` function(which enters insert mode) will actually enter terminal
mode if the current buffer is a terminal
- The `put` operator was adapted to send data to the terminal instead of
modifying the buffer directly.
- A window being resized will also trigger a terminal resize if the window
displays the terminal.
This commit pulls the some environment-variable handling functions out of
misc1.c and in to os/env.c. Previously submited as #1231, this is the start of
a patch series that does that work based on a more up-to-date master branch.
Major tasks accomplished:
- move functions and fix includes
- fix clint/clang analysis warnings
- correct documentation comments
Problem : Resource leak @ 94, 98, 102.
Diagnostic : Real issue.
Rationale : Coverity doesn't know that uv_pipe_open will save file
descriptor to close them later. So, it signals file
descriptors being leaked. This would then seem like a false
positive we can fix by teaching coverity about uv_pipe_open
through model file.
But then we realize that the above is only true if
uv_pipe_open succeeds. It it fails, then descriptors are
really being leaked, which is why this is considered a real
issue and not a false positive after all.
Resolution : Add error handling to correctly close descriptors if
uv_pipe_open fails at any point.
Add model for uv_pipe_open so that Coverity knows it will
save descriptors when no error.
Helped-by: oni-link <knil.ino@gmail.com>
Problem : Unchecked return value from library @ 91.
Diagnostic : Real issue.
Rationale : fcntl can fail, which is not being checked.
Resolution : Add corresponding error handling.
Helped-by: oni-link <knil.ino@gmail.com>
A read stream will be started before the first ex command is processed. This
stream will be used to read early user input before handling control over to the
UI module.
Which stdio stream will be used depends on which types of file descriptors are
connected, and whether the "-" argument was passed.
Even though assuming nvim is busy most times is simpler, it has a problem: A lot
of unnecessary busy_start/busy_stop notifications are sent to the UI. That's
because in the majority of scenarios almost no time is spent between
`event_poll` calls.
This restores the normal behavior which is to call busy_start only when nvim is
going to perform some task that can take a significant amount of time. Also
improve the usage of buffering in the TUI when changing the cursor state.
Switching cursor off is only necessary in two occasions:
- When redrawing to avoid terminal flickering
- When the editor is busy
The first can now be handled by the TUI, so most calls to ui_cursor_off can be
removed from the core.
So, before this commit it was only necessary to switch the cursor off to notify
the user that nvim was running some long operation. Now the cursor_{on,off}
functions have been replaced by busy_{stop,start} which can be handled in a
UI-specific way(turning the cursor off or showing a busy indicator, for
example).
To make things even more simpler, nvim is always busy except when waiting for
user input or other asynchronous events: It automatically switches to a non-busy
state when the event loop is about to be entered for more than 100 milliseconds.
`ui_busy_start` can be called when its not desired to change the busy state in
the event loop (As its now done by functions that perform blocking shell
invocations).
Compiler warns about buf always being nonnull.
buf is per function attribute always nonnull, so buf can be removed from
the assert(). But a buffer length of zero is also no problem, because it
makes uv_cwd() return a failure without writing into buf. So the
remaining length check can also be removed.
