3:34 a.m.
So I pretty regularly have frames open on multiple X displays (my
xemacs at work has frames open there as well as on my home computer),
and the biggest problem is that the DISPLAY environment variable
passed to sub-processes is always the one xemacs inherited from the
shell when it was started, rather than the one associated with the
currently selected device.
It looks like this could be fixed deep down in child_setup in
callproc.c, somewhere in the middle (in Fstart_process_internal in
process.c or maybe unix_create_process in process-unix.c) or even in
lisp by binding process-environment appropriately around the calls to
start-process-internal in start-process and call-process-internal.
Can someone give me guidance on how best to approach this?
thanks,
Greg
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
7:13 p.m.
Ar an t-ochtú lá de mí Eanair, scríobh Greg Klanderman:
So I pretty regularly have frames open on multiple X displays (my
xemacs at work has frames open there as well as on my home computer),
and the biggest problem is that the DISPLAY environment variable
passed to sub-processes is always the one xemacs inherited from the
shell when it was started, rather than the one associated with the
currently selected device.
It looks like this could be fixed deep down in child_setup in
callproc.c, somewhere in the middle (in Fstart_process_internal in
process.c or maybe unix_create_process in process-unix.c) or even in
lisp by binding process-environment appropriately around the calls to
start-process-internal in start-process and call-process-internal.
Can someone give me guidance on how best to approach this?
This should do it, in your ~/.xemacs/init.el:
(add-hook 'select-frame-hook
(lexical-let
((original-display (getenv "DISPLAY")))
#'(lambda ()
(let ((print-readably nil)
console-details)
(setq console-details
(prin1-to-string
(device-console (frame-device (selected-frame)))))
(if (string-match "#<x-console on
\"\\([^\"]+\\)"
console-details)
(setenv "DISPLAY" (match-string 1 console-details))
(setenv "DISPLAY" original-display))))))
Unfortunately, there’s no nicer way to get the string display name that a
given console thinks it’s connected to. Maybe we should change that.
--
¿Dónde estará ahora mi sobrino Yoghurtu Nghe, que tuvo que huir
precipitadamente de la aldea por culpa de la escasez de rinocerontes?
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
10:26 p.m.
On 2009-01-11, Aidan Kehoe <kehoea(a)parhasard.net> wrote:
(let ((print-readably nil)
console-details)
(setq console-details
(prin1-to-string
(device-console (frame-device (selected-frame)))))
(if (string-match "#<x-console on
\"\\([^\"]+\\)"
console-details)
(setenv "DISPLAY" (match-string 1 console-details))
(setenv "DISPLAY" original-display))))))
Unfortunately, there’s no nicer way to get the string display name
that a
given console thinks it’s connected to. Maybe we should change that.
This is what I do in my .emacs :
; this overrides make-x-device so as to store the original display
argument
; used to create it. Why? Because device-x-display doesn't exist,
documentation
; notwithstanding, and we need it to pass to external programs.
(defvar x-device-display-alist nil)
; make it robust against repeated loads of .emacs!
(when (and (fboundp 'make-x-device) (not (fboundp
'real-make-x-device)))
(add-hook 'delete-device-hook
(lambda (d) (setq x-device-display-alist
(remassoc d x-device-display-alist))))
(fset 'real-make-x-device (symbol-function 'make-x-device))
(defun make-x-device (&optional display)
"Create a new device connected to DISPLAY, recording DISPLAY in
x-device-display-alist."
(let ((d (real-make-x-device display)))
(setq x-device-display-alist (cons (cons d display)
x-device-display-alist))
d))
;; this returns nil for the initial device, but that's ok, because
;; then the functions will use DISPLAY anyway.
(defun device-x-display (device)
"Return display used to create device."
(cdr-safe (assoc device x-device-display-alist)))
)
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
3:01 p.m.
From: Julian Bradfield <jcb+xeb(a)jcbradfield.org>
Date: Sun, 11 Jan 2009 22:26:27 +0000
On 2009-01-11, Aidan Kehoe <kehoea(a)parhasard.net> wrote:
(let ((print-readably nil)
console-details)
(setq console-details
(prin1-to-string
(device-console (frame-device (selected-frame)))))
(if (string-match "#<x-console on
\"\\([^\"]+\\)"
console-details)
(setenv "DISPLAY" (match-string 1 console-details))
(setenv "DISPLAY" original-display))))))
Unfortunately, thereâs no nicer way to get the string display name
that a
given console thinks itâs connected to. Maybe we should change that.
This is what I do in my .emacs :
; this overrides make-x-device so as to store the original display
argument
; used to create it. Why? Because device-x-display doesn't exist,
documentation
; notwithstanding, and we need it to pass to external programs.
I just traced make-x-device, just to verify the display argument
1 -> make-x-device: display="localhost:10.0"
1 <- make-x-device: #<x-device on "localhost:10.0" 0x1d41f0>
device-connection returns the display of a device.
(and (eq (frame-type (selected-frame)) 'x)
(device-connection (frame-device (selected-frame))))
==>
"localhost:10.0"
(defvar x-device-display-alist nil)
; make it robust against repeated loads of .emacs!
(when (and (fboundp 'make-x-device) (not (fboundp
'real-make-x-device)))
(add-hook 'delete-device-hook
(lambda (d) (setq x-device-display-alist
(remassoc d x-device-display-alist))))
(fset 'real-make-x-device (symbol-function 'make-x-device))
(defun make-x-device (&optional display)
"Create a new device connected to DISPLAY, recording DISPLAY in
x-device-display-alist."
(let ((d (real-make-x-device display)))
(setq x-device-display-alist (cons (cons d display)
x-device-display-alist))
d))
;; this returns nil for the initial device, but that's ok, because
;; then the functions will use DISPLAY anyway.
(defun device-x-display (device)
"Return display used to create device."
(cdr-safe (assoc device x-device-display-alist)))
)
I think your code can be replaced by the following:
(defun device-x-display (device)
"Return display used to create device."
(and (eq (device-type device) 'x)
(device-connection device)))
(device-x-display (frame-device (selected-frame)))
"localhost:10.0"
-jeff
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
4:53 p.m.
Julian Bradfield writes:
device-x-display doesn't exist, documentation notwithstanding
Then that's a bug. Please report it. I would be in favor of a patch
to fix it.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta
5584
days inactive
5587
days old
4 comments
5 participants
participants (5)
-
Aidan Kehoe -
Greg Klanderman -
Jeff Mincy -
Julian Bradfield -
Stephen J. Turnbull