Calling cmd.exe in Windows follows a very different pattern from Vim.
The primary difference is that Vim does a nested call to cmd.exe, e.g.
the following call in Vim
system('echo a 2>&1')
spawns the following processes
"C:\Program Files (x86)\Vim\vim80\vimrun" -s C:\Windows\system32\cmd.exe /c (echo a 2^>^&1
^>C:\Users\dummy\AppData\Local\Temp\VIoC169.tmp 2^>^&1)
C:\Windows\system32\cmd.exe /c C:\Windows\system32\cmd.exe /c (echo a 2^>^&1
^>C:\Users\dummy\AppData\Local\Temp\VIo3C6C.tmp 2^>^&1)
C:\Windows\system32\cmd.exe /c (echo a 2>&1
>C:\Users\dummy\AppData\Local\Temp\VIo3C6C.tmp 2>&1)
The escaping with ^ is needed because cmd.exe calls itself and needs to
preserve the special metacharacters for the last call. However in nvim
no nested call is made, system('') spawns a single cmd.exe process.
Setting shellxescape to "" disables escaping with ^.
The previous default for shellxquote=( wrapped any command in
parenthesis, in Vim this is more meaningful due to the use of tempfiles
to store the output and redirection (also see &shellquote). There is
a slight benefit in having the default be empty because some expressions
that run in console will not run within parens e.g. due to unbalanced
double quotes
system('echo "a b')
Disable CommandLineToArgvW-standard quoting for cmd.exe.
libuv assumes spawned processes follow the convention expected by
CommandLineToArgvW(). But cmd.exe is non-conformant, so for cmd.exe:
- With system([]), the caller has full control (and responsibility) to
quote arguments correctly.
- With system(''), shell* options are used.
libuv quoting is disabled if argv[0] is:
- cmd.exe
- cmd
- $COMSPEC resolving to a path with filename cmd.exe
Closes#6329
References #6387
Hope this will make people using feed_command less likely: this hides bugs.
Already found at least two:
1. msgpackparse() will show internal error: hash_add() in case of duplicate
keys, though it will still work correctly. Currently silenced.
2. ttimeoutlen was spelled incorrectly, resulting in option not being set when
expected. Test was still functioning somehow though. Currently fixed.
When test/functional/eval/system_spec.lua is run on its own,
helpers.os_name() was being called before a session had been created.
This caused that describe block to fail.
Turning printargs_path into a function delays the call of
helpers.os_name() until the test is being run, which ensures a session
is available.
Do _not_ set v:shell_error on parameter validation error.
system([...]) does not invoke a shell, so this change is somewhat
questionable. But `:help v:shell_error` is sufficiently vague to allow
-1 in this case.
Closes#3529Closes#5241
In Vim,
:echo system('cat - &', 'foo')
works because for both system() and :! Vim writes input to a temp file and uses
shell syntax to redirect the file to the backgrounded `cat` (get_cmd_output()
.. make_filter_cmd()).
In Nvim,
:echo system('cat - &', 'foo')
fails because we write the input directly via pipes (shell.c:do_os_system()),
but (per POSIX[1]) backgrounded process input stream is redirected from
/dev/null (unless overridden by shell redirection; supported only by some shells
[2]), so our writes are ignored, the process exits quickly, and if we are
writing data larger than the buffer size we'll see EPIPE.
This still works:
:%w !tee > foo1358.txt &
but this does not:
:%w !tee foo1358.txt &
though it *should* (why doesn't it?) because we still do the temp file dance
in do_bang() .. do_filter().
[1] http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_09_03_02
[2] http://unix.stackexchange.com/a/71218
