bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

It would be nice if some equivalent of: 

'emacsclient -a "emacs --daemon && emacsclient -c"

would work.

That would allow one to ALWAYS use emacsclient. 

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Dan Nicolaescu <dann@ics.uci.edu> writes:

  > It would be nice if some equivalent of: 
  > 
  > 'emacsclient -a "emacs --daemon && emacsclient -c"
  > 
  > would work.
  > 
  > That would allow one to ALWAYS use emacsclient. 

Here's a quick hack that allows:

emacsclient -a "emacs --daemon && emacsclient -c" FILENAME

to work.  It is not complete yet, FILENAME does not get passed to the
"&& emacsclient"


--- emacsclient.c.~1.140.~	2008-11-03 06:30:13.000000000 -0800
+++ emacsclient.c	2008-11-18 00:13:29.000000000 -0800
@@ -630,8 +630,13 @@ fail (void)
   if (alternate_editor)
     {
       int i = optind - 1;
+      char * new_argv[] = { "sh", "-c", 0, 0 };
 
-      execvp (alternate_editor, main_argv + i);
+      new_argv[0] = "sh";
+      new_argv[1] = "-c";
+      new_argv[2] = alternate_editor;
+
+      execve ("/bin/sh", new_argv, environ);
       message (TRUE, "%s: error executing alternate editor \"%s\"\n",
 	       progname, alternate_editor);
     }


Thoughts?

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> It would be nice if some equivalent of: 
>> 'emacsclient -a "emacs --daemon && emacsclient -c"
>> would work.
>> That would allow one to ALWAYS use emacsclient. 

> Here's a quick hack that allows:
> emacsclient -a "emacs --daemon && emacsclient -c" FILENAME
> to work.  It is not complete yet, FILENAME does not get passed to the
> "&& emacsclient"

Here's an alternative:

   [ "$(emacsclient -e "'up" 2>/dev/null)" = "up" ] || emacs --daemon
   emacsclient -c FILENAME


-- Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

  > >> It would be nice if some equivalent of: 
  > >> 'emacsclient -a "emacs --daemon && emacsclient -c"
  > >> would work.
  > >> That would allow one to ALWAYS use emacsclient. 
  > 
  > > Here's a quick hack that allows:
  > > emacsclient -a "emacs --daemon && emacsclient -c" FILENAME
  > > to work.  It is not complete yet, FILENAME does not get passed to the
  > > "&& emacsclient"
  > 
  > Here's an alternative:
  > 
  >    [ "$(emacsclient -e "'up" 2>/dev/null)" = "up" ] || emacs --daemon
  >    emacsclient -c FILENAME

I've been using something similar for a long time (even before
--daemon), but it seems that it's more elegant to use a single command
that shell hackery...

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> >> It would be nice if some equivalent of: 
>> >> 'emacsclient -a "emacs --daemon && emacsclient -c"
>> >> would work.
>> >> That would allow one to ALWAYS use emacsclient. 
>> 
>> > Here's a quick hack that allows:
>> > emacsclient -a "emacs --daemon && emacsclient -c" FILENAME
>> > to work.  It is not complete yet, FILENAME does not get passed to the
>> > "&& emacsclient"
>> 
>> Here's an alternative:
>> 
>> [ "$(emacsclient -e "'up" 2>/dev/null)" = "up" ] || emacs --daemon
>> emacsclient -c FILENAME

> I've been using something similar for a long time (even before
> --daemon), but it seems that it's more elegant to use a single command
> that shell hackery...

I think both are hacks, where my hack has the advantage of requiring no
changes to the code.  If you want to change the code, than I'd rather we
do it Right.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

  > >> >> It would be nice if some equivalent of: 
  > >> >> 'emacsclient -a "emacs --daemon && emacsclient -c"
  > >> >> would work.
  > >> >> That would allow one to ALWAYS use emacsclient. 
  > >> 
  > >> > Here's a quick hack that allows:
  > >> > emacsclient -a "emacs --daemon && emacsclient -c" FILENAME
  > >> > to work.  It is not complete yet, FILENAME does not get passed to the
  > >> > "&& emacsclient"
  > >> 
  > >> Here's an alternative:
  > >> 
  > >> [ "$(emacsclient -e "'up" 2>/dev/null)" = "up" ] || emacs --daemon
  > >> emacsclient -c FILENAME
  > 
  > > I've been using something similar for a long time (even before
  > > --daemon), but it seems that it's more elegant to use a single command
  > > that shell hackery...
  > 
  > I think both are hacks, where my hack has the advantage of requiring no
  > changes to the code.  If you want to change the code, than I'd rather we
  > do it Right.

Well, now please don't tease, say what you mean by "do it Right"... 

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> I think both are hacks, where my hack has the advantage of requiring no
>> changes to the code.  If you want to change the code, than I'd rather we
>> do it Right.

> Well, now please don't tease, say what you mean by "do it Right"... 

I guess something that doesn't involve a shell would be a good start.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

  > >> I think both are hacks, where my hack has the advantage of requiring no
  > >> changes to the code.  If you want to change the code, than I'd rather we
  > >> do it Right.
  > 
  > > Well, now please don't tease, say what you mean by "do it Right"... 
  > 
  > I guess something that doesn't involve a shell would be a good start.

The the was only used because of using && in the initial example.  Any
suggestions to get around that?

How about something like:

emacsclient --start-daemon 



PS: I was thinking "do it Right" would mean: get rid of emacsclient and
do everything in emacs... 

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

> How about something like:
> emacsclient --start-daemon 

You mean "emacsclient --start-daemon-if-needed FILENAME" ?
Yes, probably something like that.  It could potentially be folded into
the alternate-editor (e.g. if alternate-editor has some special value
such as the empty string).

> PS: I was thinking "do it Right" would mean: get rid of emacsclient and
> do everything in emacs... 

;-)


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > > How about something like:
  > > emacsclient --start-daemon 
  > 
  > You mean "emacsclient --start-daemon-if-needed FILENAME" ?
  > Yes, probably something like that.  It could potentially be folded into
  > the alternate-editor (e.g. if alternate-editor has some special value
  > such as the empty string).

How does the patch below look like?
There's one refinement possible: make --start-daemon-if-needed take an
argument --start-daemon-if-needed=ARG and use it like this: emacs --daemon=ARG

Index: emacsclient.c
===================================================================
RCS file: /cvsroot/emacs/emacs/lib-src/emacsclient.c,v
retrieving revision 1.142
diff -u -3 -p -c -r1.142 emacsclient.c
cvs diff: conflicting specifications of output style
--- emacsclient.c	3 Dec 2008 04:33:44 -0000	1.142
+++ emacsclient.c	8 Dec 2008 07:52:45 -0000
@@ -150,6 +150,10 @@
 /* PID of the Emacs server process.  */
 int emacs_pid = 0;
 
+/* In case connecting to the server fails, try connecting again after
+   starting the emacs daemon.  */
+int start_daemon_if_needed = 0;
+
 void print_help_and_exit () NO_RETURN;
 
 struct option longopts[] =
@@ -162,6 +166,7 @@
   { "nw",	no_argument,       NULL, 't' },
   { "create-frame", no_argument,   NULL, 'c' },
   { "alternate-editor", required_argument, NULL, 'a' },
+  { "start-daemon-if-needed", no_argument, NULL, 'D' },
 #ifndef NO_SOCKETS_IN_FILE_SYSTEM
   { "socket-name",	required_argument, NULL, 's' },
 #endif
@@ -492,7 +497,7 @@
     {
       int opt = getopt_long_only (argc, argv,
 #ifndef NO_SOCKETS_IN_FILE_SYSTEM
-			     "VHnea:s:f:d:tc",
+			     "VHneDa:s:f:d:tc",
 #else
 			     "VHnea:f:d:tc",
 #endif
@@ -552,6 +557,10 @@
           current_frame = 0;
           break;
 
+        case 'D':
+          start_daemon_if_needed = 1;
+          break;
+
 	case 'H':
 	  print_help_and_exit ();
 	  break;
@@ -563,6 +572,12 @@
 	}
     }
 
+  if (start_daemon_if_needed && alternate_editor != NULL)
+    {
+      message (TRUE, "--alternate-editor cannot be used when using --start_daemon_if_needed\n");
+      exit (EXIT_FAILURE);
+    }
+
   /* If the -c option is used (without -t) and no --display argument
      is provided, try $DISPLAY.
      Without the -c option, we used to set `display' to $DISPLAY by
@@ -1294,7 +1309,7 @@
 #endif /* ! NO_SOCKETS_IN_FILE_SYSTEM */
 
 HSOCKET
-set_socket ()
+set_socket (int no_exit_if_error)
 {
   HSOCKET s;
 
@@ -1305,7 +1320,7 @@
   if (socket_name)
     {
       s = set_local_socket ();
-      if ((s != INVALID_SOCKET) || alternate_editor)
+      if ((s != INVALID_SOCKET) || no_exit_if_error)
 	return s;
       message (TRUE, "%s: error accessing socket \"%s\"\n",
 	       progname, socket_name);
@@ -1320,7 +1335,7 @@
   if (server_file)
     {
       s = set_tcp_socket ();
-      if ((s != INVALID_SOCKET) || alternate_editor)
+      if ((s != INVALID_SOCKET) || no_exit_if_error)
 	return s;
 
       message (TRUE, "%s: error accessing server file \"%s\"\n",
@@ -1338,7 +1353,7 @@
   /* Implicit server file.  */
   server_file = "server";
   s = set_tcp_socket ();
-  if ((s != INVALID_SOCKET) || alternate_editor)
+  if ((s != INVALID_SOCKET) || no_exit_if_error)
     return s;
 
   /* No implicit or explicit socket, and no alternate editor.  */
@@ -1353,7 +1368,7 @@
 }
 
 #ifdef WINDOWSNT
-FARPROC set_fg;  /* Pointer to AllowSetForegroundWindow.  */
+FARPROC se t_fg;  /* Pointer to AllowSetForegroundWindow.  */
 FARPROC get_wc;  /* Pointer to RealGetWindowClassA.  */
 
 BOOL CALLBACK
@@ -1416,6 +1431,8 @@
   int i, rl, needlf = 0;
   char *cwd, *str;
   char string[BUFSIZ+1];
+  int null_socket_name;
+  int null_server_file;
 
   main_argv = argv;
   progname = argv[0];
@@ -1431,9 +1448,57 @@
       exit (EXIT_FAILURE);
     }
 
-  if ((emacs_socket = set_socket ()) == INVALID_SOCKET)
-    fail ();
+  if (start_daemon_if_needed)
+    {
+      /* set_socket changes the values for socket_name and
+	 server_file, we need to reset them, if they were NULL before
+	 for the second call to set_socket.  */
+      null_socket_name = (socket_name == NULL);
+      null_server_file = (server_file == NULL);
+    }
+
+  if ((emacs_socket = set_socket (alternate_editor || start_daemon_if_needed)) == INVALID_SOCKET)
+    if (start_daemon_if_needed)
+      {
+	pid_t dpid;
+	int status;
+	pid_t p;
+
+	dpid = fork ();
 
+	if (dpid > 0)
+	  {
+	    p = waitpid (dpid, &status, WUNTRACED | WCONTINUED);
+
+	    /* Reset socket_name and server_file if they were NULL
+	       before the set_socket call.  */
+	    if (null_socket_name)
+	      socket_name = NULL;
+	    if (null_server_file)
+	      server_file = NULL;
+
+	    /* Try connecting again, the daemon should have started by
+	       now.  */
+	    message (TRUE, "daemon should have started, trying to connect again\n", dpid);
+	    if ((emacs_socket = set_socket (1)) == INVALID_SOCKET)
+	      message (TRUE, "Cannot connect even after starting the daemon\n");
+	  }
+	else if (dpid < 0)
+	  {
+	    fprintf (stderr, "Cannot fork!\n");
+	    exit (1);
+	  }
+	else
+	  {
+	    char *const d_argv[] = {"emacs", "--daemon", 0 };
+
+	    /* Start the daemon. */
+	    execvp ("emacs", d_argv);
+	    message (TRUE, "%s: error starting emacs daemon\n", progname);
+	  }
+      }
+    else
+      fail ();
 
   cwd = get_current_dir_name ();
   if (cwd == 0)

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> > How about something like:
>> > emacsclient --start-daemon
>> You mean "emacsclient --start-daemon-if-needed FILENAME" ?
>> Yes, probably something like that.  It could potentially be folded into
>> the alternate-editor (e.g. if alternate-editor has some special value
>> such as the empty string).
> How does the patch below look like?

I'd prefer to combine it with --alternate-editor since the two args are
mutually exclusive.  Have you tried it and it caused problems?

> There's one refinement possible: make --start-daemon-if-needed take an
> argument --start-daemon-if-needed=ARG and use it like this:
> emacs --daemon=ARG

I think this ARG should automatically be taken from the
usual --socket-name argument.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > >> > How about something like:
  > >> > emacsclient --start-daemon
  > >> You mean "emacsclient --start-daemon-if-needed FILENAME" ?
  > >> Yes, probably something like that.  It could potentially be folded into
  > >> the alternate-editor (e.g. if alternate-editor has some special value
  > >> such as the empty string).
  > > How does the patch below look like?
  > 
  > I'd prefer to combine it with --alternate-editor since the two args are
  > mutually exclusive.  Have you tried it and it caused problems?

I didn't quite get what you said... It should probably work, modulo ... 

  > > There's one refinement possible: make --start-daemon-if-needed take an
  > > argument --start-daemon-if-needed=ARG and use it like this:
  > > emacs --daemon=ARG
  > 
  > I think this ARG should automatically be taken from the
  > usual --socket-name argument.

... this: That won't work in the future when we extend --daemon to be able to
specify a TCP port for example: --daemon=FILENAME:TCP_PORT

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> >> > How about something like:
>> >> > emacsclient --start-daemon
>> >> You mean "emacsclient --start-daemon-if-needed FILENAME" ?
>> >> Yes, probably something like that.  It could potentially be folded into
>> >> the alternate-editor (e.g. if alternate-editor has some special value
>> >> such as the empty string).
>> > How does the patch below look like?
>> 
>> I'd prefer to combine it with --alternate-editor since the two args are
>> mutually exclusive.  Have you tried it and it caused problems?

> I didn't quite get what you said... It should probably work, modulo ... 

>> > There's one refinement possible: make --start-daemon-if-needed take an
>> > argument --start-daemon-if-needed=ARG and use it like this:
>> > emacs --daemon=ARG
>> 
>> I think this ARG should automatically be taken from the
>> usual --socket-name argument.

> ... this: That won't work in the future when we extend --daemon to be able to
> specify a TCP port for example: --daemon=FILENAME:TCP_PORT

No need to worry about that: I've already explicitly rejected a tcp-port
option for server.el in the past, so I'll be happy to give other such
rejections in the future.


        Stefan


PS: The magical value for --alternate-editory should hopefully also work
in $ALTERNATE_EDITOR.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > >> >> > How about something like:
  > >> >> > emacsclient --start-daemon
  > >> >> You mean "emacsclient --start-daemon-if-needed FILENAME" ?
  > >> >> Yes, probably something like that.  It could potentially be folded into
  > >> >> the alternate-editor (e.g. if alternate-editor has some special value
  > >> >> such as the empty string).
  > >> > How does the patch below look like?
  > >> 
  > >> I'd prefer to combine it with --alternate-editor since the two args are
  > >> mutually exclusive.  Have you tried it and it caused problems?
  > 
  > > I didn't quite get what you said... It should probably work, modulo ... 
  > 
  > >> > There's one refinement possible: make --start-daemon-if-needed take an
  > >> > argument --start-daemon-if-needed=ARG and use it like this:
  > >> > emacs --daemon=ARG
  > >> 
  > >> I think this ARG should automatically be taken from the
  > >> usual --socket-name argument.
  > 
  > > ... this: That won't work in the future when we extend --daemon to be able to
  > > specify a TCP port for example: --daemon=FILENAME:TCP_PORT
  > 
  > No need to worry about that: I've already explicitly rejected a tcp-port
  > option for server.el in the past, so I'll be happy to give other such
  > rejections in the future.

Hmm, you said the reverse here: http://article.gmane.org/gmane.emacs.devel/103350

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> No need to worry about that: I've already explicitly rejected a tcp-port
>> option for server.el in the past, so I'll be happy to give other such
>> rejections in the future.

> Hmm, you said the reverse here:
> http://article.gmane.org/gmane.emacs.devel/103350

Indeed.  Consistency is the hgolm, isn't it?
In any case, I think it's more important to handle the alternate-editor
issue than the tcp port issue.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

  > >> No need to worry about that: I've already explicitly rejected a tcp-port
  > >> option for server.el in the past, so I'll be happy to give other such
  > >> rejections in the future.
  > 
  > > Hmm, you said the reverse here:
  > > http://article.gmane.org/gmane.emacs.devel/103350
  > 
  > Indeed.  Consistency is the hgolm, isn't it?

hgolm?

Updated patch that uses -a ""  or --alternate-editor= or
ALTERNATE_EDITOR="":

Index: emacsclient.c
===================================================================
RCS file: /cvsroot/emacs/emacs/lib-src/emacsclient.c,v
retrieving revision 1.142
diff -u -3 -p -u -p -r1.142 emacsclient.c
--- emacsclient.c	3 Dec 2008 04:33:44 -0000	1.142
+++ emacsclient.c	9 Dec 2008 18:53:51 -0000
@@ -1294,7 +1294,7 @@ To start the server in Emacs, type \"M-x
 #endif /* ! NO_SOCKETS_IN_FILE_SYSTEM */
 
 HSOCKET
-set_socket ()
+set_socket (int no_exit_if_error)
 {
   HSOCKET s;
 
@@ -1305,7 +1305,7 @@ set_socket ()
   if (socket_name)
     {
       s = set_local_socket ();
-      if ((s != INVALID_SOCKET) || alternate_editor)
+      if ((s != INVALID_SOCKET) || no_exit_if_error)
 	return s;
       message (TRUE, "%s: error accessing socket \"%s\"\n",
 	       progname, socket_name);
@@ -1320,7 +1320,7 @@ set_socket ()
   if (server_file)
     {
       s = set_tcp_socket ();
-      if ((s != INVALID_SOCKET) || alternate_editor)
+      if ((s != INVALID_SOCKET) || no_exit_if_error)
 	return s;
 
       message (TRUE, "%s: error accessing server file \"%s\"\n",
@@ -1338,7 +1338,7 @@ set_socket ()
   /* Implicit server file.  */
   server_file = "server";
   s = set_tcp_socket ();
-  if ((s != INVALID_SOCKET) || alternate_editor)
+  if ((s != INVALID_SOCKET) || no_exit_if_error)
     return s;
 
   /* No implicit or explicit socket, and no alternate editor.  */
@@ -1416,6 +1416,7 @@ main (argc, argv)
   int i, rl, needlf = 0;
   char *cwd, *str;
   char string[BUFSIZ+1];
+  int null_socket_name, null_server_file, start_daemon_if_needed;
 
   main_argv = argv;
   progname = argv[0];
@@ -1431,9 +1432,68 @@ main (argc, argv)
       exit (EXIT_FAILURE);
     }
 
-  if ((emacs_socket = set_socket ()) == INVALID_SOCKET)
-    fail ();
-
+  start_daemon_if_needed = (alternate_editor
+			    && (alternate_editor[0] == '\0'));
+  if (start_daemon_if_needed)
+    {
+      /* set_socket changes the values for socket_name and
+	 server_file, we need to reset them, if they were NULL before
+	 for the second call to set_socket.  */
+      null_socket_name = (socket_name == NULL);
+      null_server_file = (server_file == NULL);
+    }
+
+  if ((emacs_socket = set_socket (alternate_editor
+				  || start_daemon_if_needed)) == INVALID_SOCKET)
+    if (start_daemon_if_needed)
+      {
+	pid_t dpid;
+	int status;
+	pid_t p;
+
+	/* Reset socket_name and server_file if they were NULL
+	   before the set_socket call.  */
+	if (null_socket_name)
+	  socket_name = NULL;
+	if (null_server_file)
+	  server_file = NULL;
+
+	dpid = fork ();
+
+	if (dpid > 0)
+	  {
+	    p = waitpid (dpid, &status, WUNTRACED | WCONTINUED);
+
+	    /* Try connecting again, the daemon should have started by
+	       now.  */
+	    message (TRUE, "daemon should have started, trying to connect again\n", dpid);
+	    if ((emacs_socket = set_socket (1)) == INVALID_SOCKET)
+	      message (TRUE, "Cannot connect even after starting the daemon\n");
+	  }
+	else if (dpid < 0)
+	  {
+	    fprintf (stderr, "Cannot fork!\n");
+	    exit (1);
+	  }
+	else
+	  {
+	    char *d_argv[] = {"emacs", "--daemon", 0 };
+	    if (!null_socket_name)
+	      {
+		/* Pass  --daemon=socket_name as argument.  */
+		char *deq = "--daemon=";
+		char *daemon_arg = alloca (strlen (deq)
+					   + strlen (socket_name) + 1);
+		strcpy (daemon_arg, deq);
+		strcat (daemon_arg, socket_name);
+		d_argv[1] = daemon_arg;
+	      }
+	    execvp ("emacs", d_argv);
+	    message (TRUE, "%s: error starting emacs daemon\n", progname);
+	  }
+      }
+    else
+      fail ();
 
   cwd = get_current_dir_name ();
   if (cwd == 0)

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Tue, Dec 9, 2008 at 19:58, Dan Nicolaescu <dann@ics.uci.edu> wrote:

>  > Indeed.  Consistency is the hgolm, isn't it?
>
> hgolm?

http://www.bartleby.com/59/3/foolishconsi.html

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

>> >> No need to worry about that: I've already explicitly rejected a tcp-port
>> >> option for server.el in the past, so I'll be happy to give other such
>> >> rejections in the future.
>> > Hmm, you said the reverse here:
>> > http://article.gmane.org/gmane.emacs.devel/103350
>> Indeed.  Consistency is the hgolm, isn't it?
> hgolm?

http://www.bartleby.com/59/3/foolishconsi.html

> Updated patch that uses -a ""  or --alternate-editor= or
> ALTERNATE_EDITOR="":

Looks OK, except that the whole body of the if, where you do the fork
etc.. deserves its own function.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

  > >> >> No need to worry about that: I've already explicitly rejected a tcp-port
  > >> >> option for server.el in the past, so I'll be happy to give other such
  > >> >> rejections in the future.
  > >> > Hmm, you said the reverse here:
  > >> > http://article.gmane.org/gmane.emacs.devel/103350
  > >> Indeed.  Consistency is the hgolm, isn't it?
  > > hgolm?
  > 
  > http://www.bartleby.com/59/3/foolishconsi.html
  > 
  > > Updated patch that uses -a ""  or --alternate-editor= or
  > > ALTERNATE_EDITOR="":
  > 
  > Looks OK, except that the whole body of the if, where you do the fork
  > etc.. deserves its own function.

Thanks. Checked in with that change.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Wed, Dec 10, 2008 at 08:58, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> Thanks. Checked in with that change.

I've checked in the minimal change that allows emacsclient to build on
Windows again.

Feel free to adapt to your tastes, but please, don't delete the
EMACS_DAEMON define.

  Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

"Juanma Barranquero" <lekktu@gmail.com> writes:

  > On Wed, Dec 10, 2008 at 08:58, Dan Nicolaescu <dann@ics.uci.edu> wrote:
  > 
  > > Thanks. Checked in with that change.
  > 
  > I've checked in the minimal change that allows emacsclient to build on
  > Windows again.
  > 
  > Feel free to adapt to your tastes, but please, don't delete the
  > EMACS_DAEMON define.

As you well know my taste is exactly not to add such #defines.  There's
not such #define in the daemon implementation in src, the new macro in
undocumented, and your move was just designed as a provocation.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Wed, Dec 10, 2008 at 16:03, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> As you well know my taste is exactly not to add such #defines.

As you well know, some of us prefer such defines, when needed.

> There's
> not such #define in the daemon implementation in src,

But it is neeeded. What you just committed does *not* work on Windows.

> the new macro in
> undocumented,

That can easily be fixed by documenting.

> and your move was just designed as a provocation.

No, it was not; it's just that I like my emacsclient working. But
curiously I've got the same feeling from your action: that is was just
a provocation. What do you want for us to do, to engage in a pointless
commit war (because, IIRC, emacsclient.c is not yours only to decide).

Do not remove Windows changes unless you're providing an equivalent,
or better, fix. So please, revert your change or fix it so I *don't*
need to use an alternate server.

  C:\emacs> emacsclient --alternate-editor=c:\emacs\bin\emacs.exe my-file.txt
  emacsclient: connect: No se ha podido establecer conexi¾n ya que el
equipo de destino ha
  denegado activamente dicha conexi¾n

The message says: "could not establish a connection because the
destination computer actively rejected the connection", more or less.

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

"Juanma Barranquero" <lekktu@gmail.com> writes:

  > On Wed, Dec 10, 2008 at 16:03, Dan Nicolaescu <dann@ics.uci.edu> wrote:
  > 
  > 
  > > There's
  > > not such #define in the daemon implementation in src,
  > 
  > But it is neeeded. 

That's a separate discussion that I am not interested in.  When such a
macro is added, this can be revisited and be made consistent.

  > What you just committed does *not* work on Windows.

There was a bug that I fixed, not related to any new macro.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Wed, Dec 10, 2008 at 16:29, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> That's a separate discussion that I am not interested in.

There's quite a difference between "I'm not interested in discussing
this" and "I'm going to revert your changes because I don't like what
you did, even if I know that you do".

> When such a
> macro is added, this can be revisited and be made consistent.

Perhaps you've forgotten that the macro was added. And deleted. By you.

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

"Juanma Barranquero" <lekktu@gmail.com> writes:

  > On Wed, Dec 10, 2008 at 16:29, Dan Nicolaescu <dann@ics.uci.edu> wrote:
  > 
  > > That's a separate discussion that I am not interested in.
  > 
  > There's quite a difference between "I'm not interested in discussing
  > this" and "I'm going to revert your changes because I don't like what
  > you did, even if I know that you do".

Omitting quotes to modify the meaning is not very productive:

The sequence is:
1.
Dan> There's
Dan> not such #define in the daemon implementation in src,

2.
Juanma> But it is neeeded. What you just committed does *not* work on Windows.

3.
Dan> That's a separate discussion that I am not interested in.
Dan> When such a
Dan> macro is added, this can be revisited and be made consistent.

Can you please STOP this?

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Wed, Dec 10, 2008 at 16:43, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> Omitting quotes to modify the meaning is not very productive:

Then don't do it.

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

> As you well know my taste is exactly not to add such #defines.  There's
> not such #define in the daemon implementation in src, the new macro in
> undocumented, and your move was just designed as a provocation.

Please, let's calm down.

I generally agree that it's better to use "#ifdef <FEATURE>" than
"#ifdef <PLATFORM>" (tho it always depends on the specifics), so to
a large extent I prefer Juanma's code in this case.  But in any case,
when it's a question of taste, the one who wrote the code (Juanma in
this case, even if it touched the code you wrote) gets to choose.
Only ever change someone else's code when it's uncontroversial, or only
after a discussion here concluded it's better to change it.  In this
case, it's clearly controversial.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > > As you well know my taste is exactly not to add such #defines.  There's
  > > not such #define in the daemon implementation in src, the new macro in
  > > undocumented, and your move was just designed as a provocation.
  > 
  > Please, let's calm down.
  > 
  > I generally agree that it's better to use "#ifdef <FEATURE>" than
  > "#ifdef <PLATFORM>" (tho it always depends on the specifics), so to
  > a large extent I prefer Juanma's code in this case.  But in any case,
  > when it's a question of taste, the one who wrote the code (Juanma in
  > this case, even if it touched the code you wrote) gets to choose.
  > Only ever change someone else's code when it's uncontroversial, or only
  > after a discussion here concluded it's better to change it.  In this
  > case, it's clearly controversial.

I think you got it wrong here.  I wrote the code, Juanma just sprinkled
a few #ifdefs to simply disable it on windows, nothing else.  He knew
clearly well what my taste is.  Given that I get to fix bugs in the
code, I prefer for it to be done in the way I think it's right.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Thu, Dec 11, 2008 at 18:08, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> I think you got it wrong here.  I wrote the code

Really? All of emacsclient.c, or it's only the parts you touch that turn yours?

> Juanma just sprinkled
> a few #ifdefs to simply disable it on windows, nothing else.

Your revisionism won't change the fact that I didn't "disable things
on Windows", I made it work *twice* where you carelessly broke it (on
Windows and, apparently, also on GNU/Linux).

> Given that I get to fix bugs in the
> code, I prefer for it to be done in the way I think it's right.

That's the very reason why I did it my way: because I was fixing bugs.
The ones you introduced.

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

> I think you got it wrong here.  I wrote the code, Juanma just sprinkled
> a few #ifdefs to simply disable it on windows, nothing else.  He knew
> clearly well what my taste is.  Given that I get to fix bugs in the
> code, I prefer for it to be done in the way I think it's right.

I know you disagree, but he wrote the ifdefs, so the ifdefs are
his code.


        Stefan

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Thu, Dec 11, 2008 at 20:06, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> I know you disagree, but he wrote the ifdefs, so the ifdefs are
> his code.

OTOH, we could apply his reasoning to the TCP code in emacsclient.c (I
wrote most of it, after all), so I get rights to change everything Dan
makes in that code...

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > > I think you got it wrong here.  I wrote the code, Juanma just sprinkled
  > > a few #ifdefs to simply disable it on windows, nothing else.  He knew
  > > clearly well what my taste is.  Given that I get to fix bugs in the
  > > code, I prefer for it to be done in the way I think it's right.
  > 
  > I know you disagree, but he wrote the ifdefs, so the ifdefs are
  > his code.

By that reasoning, should anyone feel free to change your code in a way
we know you dislike and that would be fine?  And you'd just let that
stand?  I don't think so.

I would have done the changes to disable the code on windows, but Juanma
was just quicker (so thanks, but no thanks) and intentionally chose to
be confrontational, as seen from all his extremely aggressive followup
emails and continuous misinterpretation and badgering.  I don't like to
tolerate bullies like that.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Thu, Dec 11, 2008 at 20:28, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> intentionally chose to
> be confrontational

Yes, of course. I usually say "please" when I try to be confrontational.

> as seen from all his extremely aggressive followup
> emails and continuous misinterpretation and badgering.  I don't like to
> tolerate bullies like that.

The pot, calling the kettle black..

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Dan Nicolaescu]

"Juanma Barranquero" <lekktu@gmail.com> writes:

  > On Thu, Dec 11, 2008 at 20:28, Dan Nicolaescu <dann@ics.uci.edu> wrote:
  > 
  > > intentionally chose to
  > > be confrontational
  > 
  > Yes, of course. I usually say "please" when I try to be confrontational.

Your actions speak by themselves: you had no contribution to that code,
no contribution to thinking about the problem, how to solve it,
discussing it, coding, getting it approved.
You also have no interest in improving the code, just getting it out of
the way.
You knew exactly what my opinion was on adding extra #defines, but
despite all the above, you chose ignore that opinion, not even ask, and
barrel through and do it that way you wanted.
That's confrontational, and you've been that way in ALL your followup
messages.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Juanma Barranquero]

On Thu, Dec 11, 2008 at 21:39, Dan Nicolaescu <dann@ics.uci.edu> wrote:

> Your actions speak by themselves: you had no contribution to that code,
> no contribution to thinking about the problem, how to solve it,
> discussing it, coding, getting it approved.

Things that you're obviously doing a lot.

> You also have no interest in improving the code, just getting it out of
> the way.

I moved code out of the way. Your "fix" to my fix was, in both cases,
leaving the same code "out of the way". In one case you changed the
define, and in the other you removed the code I had #ifdef'd out. So I
was not that misguided, it seems.

> That's confrontational, and you've been that way in ALL your followup
> messages.

Kettle, pot, black again.

    Juanma

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Stefan Monnier]

This discussion is pointless and hence over.


        Stefan


>>>>> "Dan" == Dan Nicolaescu <dann@ics.uci.edu> writes:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> > I think you got it wrong here.  I wrote the code, Juanma just sprinkled
>> > a few #ifdefs to simply disable it on windows, nothing else.  He knew
>> > clearly well what my taste is.  Given that I get to fix bugs in the
>> > code, I prefer for it to be done in the way I think it's right.
>> 
>> I know you disagree, but he wrote the ifdefs, so the ifdefs are
>> his code.

> By that reasoning, should anyone feel free to change your code in a way
> we know you dislike and that would be fine?  And you'd just let that
> stand?  I don't think so.

> I would have done the changes to disable the code on windows, but Juanma
> was just quicker (so thanks, but no thanks) and intentionally chose to
> be confrontational, as seen from all his extremely aggressive followup
> emails and continuous misinterpretation and badgering.  I don't like to
> tolerate bullies like that.

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Andreas Schwab]

Dan Nicolaescu <dann@ics.uci.edu> writes:

> Well, now please don't tease, say what you mean by "do it Right"... 

Use a shell script.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'

[Eli Zaretskii]

> Date: Tue, 18 Nov 2008 00:22:33 -0800 (PST)
> From: Dan Nicolaescu <dann@ics.uci.edu>
> Cc: 
> 
> --- emacsclient.c.~1.140.~	2008-11-03 06:30:13.000000000 -0800
> +++ emacsclient.c	2008-11-18 00:13:29.000000000 -0800
> @@ -630,8 +630,13 @@ fail (void)
>    if (alternate_editor)
>      {
>        int i = optind - 1;
> +      char * new_argv[] = { "sh", "-c", 0, 0 };
>  
> -      execvp (alternate_editor, main_argv + i);
> +      new_argv[0] = "sh";
> +      new_argv[1] = "-c";
> +      new_argv[2] = alternate_editor;
> +
> +      execve ("/bin/sh", new_argv, environ);
>        message (TRUE, "%s: error executing alternate editor \"%s\"\n",
>  	       progname, alternate_editor);
>      }
> 
> 
> Thoughts?

Won't work on platforms that don't have /bin/sh.

bug#1298: marked as done (allow 'emacsclient -a "emacs --daemon && emacsclient -c"')

[Emacs bug Tracking System]

[-- Attachment #1: Type: text/plain, Size: 912 bytes --]


Your message dated Thu, 11 Dec 2008 13:23:16 -0800 (PST)
with message-id <200812112123.mBBLNGpI001493@mothra.ics.uci.edu>
and subject line Re: bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'
has caused the Emacs bug report #1298,
regarding allow 'emacsclient -a "emacs --daemon && emacsclient -c"'
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact don@donarmstrong.com
immediately.)


-- 
1298: http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=1298
Emacs Bug Tracking System
Contact don@donarmstrong.com with problems

[-- Attachment #2: Type: message/rfc822, Size: 2563 bytes --]

From: Dan Nicolaescu <dann@ics.uci.edu>
To: bug-gnu-emacs <bug-gnu-emacs@gnu.org>
Subject: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'
Date: Sun, 2 Nov 2008 10:18:15 -0800 (PST)
Message-ID: <200811021818.mA2IIFMq028772@mothra.ics.uci.edu>


It would be nice if some equivalent of: 

'emacsclient -a "emacs --daemon && emacsclient -c"

would work.

That would allow one to ALWAYS use emacsclient. 




[-- Attachment #3: Type: message/rfc822, Size: 2701 bytes --]

From: Dan Nicolaescu <dann@ics.uci.edu>
To: 1298-done@emacsbugs.donarmstrong.com
Subject: Re: bug#1298: allow 'emacsclient -a "emacs --daemon && emacsclient -c"'
Date: Thu, 11 Dec 2008 13:23:16 -0800 (PST)
Message-ID: <200812112123.mBBLNGpI001493@mothra.ics.uci.edu>

Fixed.