remote.txt For Vim version 9.1. Last change: 2025 Aug 22 VIM REFERENCE MANUAL by Bram Moolenaar Vim client-server communication client-server 1. Common functionality clientserver 2. X11 specific items x11-clientserver 3. MS-Windows specific items w32-clientserver 4. Socket server specific items socketserver-clientserver ============================================================================== 1. Common functionality clientserver When compiled with the +clientserver option, Vim can act as a command server. It accepts messages from a client and executes them. At the same time, Vim can function as a client and send commands to a Vim server. The following command line arguments are available: argument meaning --remote [+{cmd}] {file} ... --remote Open the file list in a remote Vim. When there is no Vim server, execute locally. There is one optional init command: +{cmd}. This must be an Ex command that can be followed by "|". The rest of the command line is taken as the file list. Thus any non-file arguments must come before this. You cannot edit stdin this way --. The remote Vim is raised. If you don't want this use vim --remote-send "<C-\><C-N>:n filename<CR>" --remote-silent [+{cmd}] {file} ... --remote-silent As above, but don't complain if there is no server and the file is edited locally. --remote-wait [+{cmd}] {file} ... --remote-wait As --remote, but wait for files to complete (unload) in remote Vim. --remote-wait-silent [+{cmd}] {file} ... --remote-wait-silent As --remote-wait, but don't complain if there is no server. --remote-tab --remote-tab Like --remote but open each file in a new tabpage. --remote-tab-silent --remote-tab-silent Like --remote-silent but open each file in a new tabpage. --remote-tab-wait --remote-tab-wait Like --remote-wait but open each file in a new tabpage. --remote-tab-wait-silent --remote-tab-wait-silent Like --remote-wait-silent but open each file in a new tabpage. --servername --servername {name} Become the server {name}. When used together with one of the --remote commands: connect to server {name} instead of the default (see below). The name used will be uppercase. If using the socketserver, you can specify a path, see socketserver-name for more details. --remote-send --remote-send {keys} Send {keys} to server and exit. The {keys} are not mapped. Special key names are recognized, e.g., "<CR>" results in a CR character. --remote-expr --remote-expr {expr} Evaluate {expr} in server and print the result on stdout. --serverlist --serverlist Output a list of server names. --clientserver --clientserver {method} Use the specified method {method} as the backend for clientserver functionality. Can either be "socket" or "x11". {only available when compiled with both +X11 and +socketserver features} Examples Edit "file.txt" in an already running GVIM server: gvim --remote file.txt Edit "file.txt" in an already running server called FOOBAR: gvim --servername FOOBAR --remote file.txt Edit "file.txt" in server "FILES" if it exists, become server "FILES" otherwise: gvim --servername FILES --remote-silent file.txt This doesn't work, all arguments after --remote will be used as file names: gvim --remote --servername FOOBAR file.txt Edit file "+foo" in a remote server (note the use of "./" to avoid the special meaning of the leading plus): vim --remote ./+foo Tell the remote server "BLA" to write all files and exit: vim --servername BLA --remote-send '<C-\><C-N>:wqa<CR>' SERVER NAME client-server-name By default Vim will try to register the name under which it was invoked (gvim, egvim ...). This can be overridden with the --servername argument. If the specified name is not available, a postfix is applied until a free name is encountered, i.e. "gvim1" for the second invocation of gvim on a particular X-server. The resulting name is available in the servername builtin variable v:servername. The case of the server name is ignored, thus "gvim" and "GVIM" are considered equal. Note if a socket server is being used, there are some differences, see socketserver-differences. When Vim is invoked with --remote, --remote-wait or --remote-send it will try to locate the server name determined by the invocation name and --servername argument as described above. If an exact match is not available, the first server with the number postfix will be used. If a name with the number postfix is specified with the --servername argument, it must match exactly. If no server can be located and --remote or --remote-wait was used, Vim will start up according to the rest of the command line and do the editing by itself. This way it is not necessary to know whether gvim is already started when sending command to it. The --serverlist argument will cause Vim to print a list of registered command servers on the standard output (stdout) and exit. If a socket server is being used, there are caveats, see socketserver-differences. {server} The {server} argument is used by several functions. When this is an empty string then on Unix the default server name is used, which is "GVIM". On MS-Windows an empty string does not work. Win32 Note: Making the Vim server go to the foreground doesn't always work, because MS-Windows doesn't allow it. The client will move the server to the foreground when using the --remote or --remote-wait argument and the server name starts with "g". REMOTE EDITING The --remote argument will cause a :drop command to be constructed from the rest of the command line and sent as described above. The --remote-wait argument does the same thing and additionally sets up to wait for each of the files to have been edited. This uses the BufUnload event, thus as soon as a file has been unloaded, Vim assumes you are done editing it. Note that the --remote and --remote-wait arguments will consume the rest of the command line. I.e. all remaining arguments will be regarded as filenames. You can not put options there! FUNCTIONS E240 E573 There are a number of Vim functions for scripting the command server. See the description in builtin.txt or use CTRL-] on the function name to jump to the full explanation. synopsis explanation remote_startserver( name) run a server remote_expr( server, string, idvar) send expression remote_send( server, string, idvar) send key sequence serverlist() get a list of available servers remote_peek( serverid, retvar) check for reply string remote_read( serverid) read reply string server2client( serverid, string) send reply string remote_foreground( server) bring server to the front See also the explanation of CTRL-\_CTRL-N. Very useful as a leading key sequence. The {serverid} for server2client() can be obtained with expand("<client>") ============================================================================== 2. X11 specific items x11-clientserver E247 E248 E251 E258 E277 The communication between client and server goes through the X server. The display of the Vim server must be specified. The usual protection of the X server is used, you must be able to open a window on the X server for the communication to work. It is possible to communicate between different systems. By default, a GUI Vim will register a name on the X-server by which it can be addressed for subsequent execution of injected strings. Vim can also act as a client and send strings to other instances of Vim on the same X11 display. When an X11 GUI Vim (gvim) is started, it will try to register a send-server name on the 'VimRegistry' property on the root window. A non GUI Vim with access to the X11 display (xterm-clipboard enabled), can also act as a command server if a server name is explicitly given with the --servername argument, or when Vim was built with the +autoservername feature. An empty --servername argument will cause the command server to be disabled. To send commands to a Vim server from another application, read the source file src/if_xcmdsrv.c, it contains some hints about the protocol used. ============================================================================== 3. Win32 specific items w32-clientserver Every Win32 Vim can work as a server, also in the console. You do not need a version compiled with OLE. Windows messages are used, this works on any version of MS-Windows. But only communication within one system is possible. Since MS-Windows messages are used, any other application should be able to communicate with a Vim server. An alternative is using the OLE functionality ole-interface. When using gvim, the --remote-wait only works properly this way: start /w gvim --remote-wait file.txt ============================================================================== 4. Socket server specific items socketserver-clientserver E1563 E1564 E1565 E1566 E1567 The communication between client and server is done using Unix domain sockets. These sockets are either placed in these directories in the following order of availability: 1. "$XDG_RUTIME_DIR/vim" if $XDG_RUNTIME_DIR is set in the environment. 2. "$TMPDIR/vim-[uid]", where "[uid]" is the uid of the user. This directory will have the access permissions set to 700 so only the user can read or write from/to it. If $TMPDIR is not set, "/tmp" is used. socketserver-name When specifying the server id/name, it can be taken as a generic name or an absolute or relative path. If the server id starts with either a "/" (absolute) or "./" | "../" (relative), then it is taken as path to the socket. Otherwise the server id will be the filename of the socket which will be placed in the above common directories. Note that a server id/name can only contain slashes "/" if it is taken as a path, so names such as "abc/dir" will be invalid. Socket server functionality is available in both GTK GUI and terminal versions of Vim. Unless Vim is compiled with +autoservername feature, the socket server will have to started explicitly, just like X11, even in the GUI. If Vim crashes or does not exit cleanly, the socket server will not remove the socket file and it will be left around. This is generally not a problem, because if a socket name is taken, Vim checks if the socket in its place is dead (not attached to any process), and can replace it instead of finding a new name. To send commands to a Vim socket server from another application, read the source file src/os_unix.c, there is detailed description of the protocol used. socketserver-differences Most of the functionality is the same as X11, however unlike X11, where the client does not need to be a server in order to communicate with another server, the socket server requires the server to be running even as a client. The exception is serverlist() or the --serverlist argument, which does not require the server to be running. Additionally, the server id or client id will not be a number like X11 or MS-Windows (shown in hex representation), instead it is the absolute path to the socket. This can be seen via the v:servername variable. The --serverlist argument will act just like X11, however it only checks the given common directories above. If a custom path is used for a socket, it will not be detected, such as a path either not in $XDG_RUNTIME_DIR or <$TMPDIR or /tmp>/vim of the --serverlist Vim process. If you have both +socketserver and +X11 compiled, you will need to add --clientserver set to "socket" in combination with --serverlist to list the available servers. You cannot list both types of backends in one command. socketserver-x11 If Vim is compiled with both +X11 and +socketserver, then deciding which backend to use is done at startup time, via the --clientserver argument. By default if it is not specified, then X11 will be used. A Vim instance using a socket server cannot communicate with one using X11. vim:tw=78:sw=4:ts=8:noet:ft=help:norl: