bug#3128: Bug#3128: emacsclient 22.3 race condition when launching emacs is required

[Terence Dineen <terence.dineen@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2239 bytes --]

 Glenn Morris wrote:

Terence Dineen wrote:



 $ echo $ALTERNATE_EDITOR
/sandbox/bin/emacs
$ emacsclient foo
emacsclient: connect: Connection refused

The newly launched emacs has a buffer open on foo but not as a server
buffer.

emacsclient is waiting for a done message from emacs and emacs is acting as
though it was launched by emacsclient using the --no-wait option.


 I don't really understand what you are saying the problem is, but if
you get the 23.0.95 pretest from
ftp://alpha.gnu.org/gnu/emacs/pretest/emacs-23.0.95.tar.gz

I think that

emacsclient -c -a "" foo

might do what you want it to do.


 Sorry, I see that my report was too terse.

I have built 23,0.95 and run the command you suggested.  It does behave as I
expected.

What I was trying to say in my bug report against 22.3 was:

If I have an emacs process running already and execute "emacsclient foo" the
result is that a new buffer appears in the emacs process as a server editing
buffer i.e. it has a message "When done with a buffer, type C-x #".  The
emacsclient invocation waits until the server buffer is dismissed.

On the other hand, if I do not have an emacs process running already before
executing "emacsclient foo" the result is message from emacsclient

    emacsclient: connect: Connection refused

an emacs process is started and it has a buffer on foo.  However that buffer
is not a server editing buffer (C-x # does nothing) and the emacsclient
process does not terminate until the emacs process it launched is
terminated.  Maybe I am missing something, but my expection was that
emacsclient would behave roughly as follows when there is no emacs process
alreadt running:

1) notice the absence of a running emacs
2) use the -a option or the ALTERNATIVE_EDITOR to find another image to
launch
3) launch it and wait for it to open the appropriate socket
4) start doing whatever it would have done to tell the now running emacs to
open a server editing buffer on foo if it had found a running emacs in the
first place.

I guess I can now see that emacsclient can't know that the alternate editor
is even emacs so it can't use the same protocol as it would have had it
found a running emacs server in the first place.

Sorry for the distraction.

[-- Attachment #2: Type: text/html, Size: 2703 bytes --]