Development list for the Stump WM.

headers Ben Spencer | 2 Jan 2010 22:53

I've now pushed this patch.  Let me know if it causes any issues.

Ben

---
 wrappers.lisp |   26 ++++++++++++--------------
 1 files changed, 12 insertions(+), 14 deletions(-)

diff --git a/wrappers.lisp b/wrappers.lisp
index 658b1c5..89f9626 100644
--- a/wrappers.lisp
+++ b/wrappers.lisp
 <at>  <at>  -212,26 +212,24  <at>  <at> 

 (defun utf8-to-string (octets)
   "Convert the list of octets to a string."
-  #+sbcl (handler-bind
-             ((sb-impl::octet-decoding-error #'(lambda (c) 
-                                                 (declare (ignore c))
-                                                 (invoke-restart 'use-value "?"))))
-           (sb-ext:octets-to-string 
-            (coerce octets '(vector (unsigned-byte 8)))
-            :external-format :utf-8))
-  #+clisp (ext:convert-string-from-bytes (coerce octets '(vector (unsigned-byte 8)))
-                                         charset:utf-8)
-  #-(or sbcl clisp)
-  (map 'string #'code-char octets))
+  (let ((octets (coerce octets '(vector (unsigned-byte 8)))))
+    #+ccl (ccl:decode-string-from-octets octets :external-format :utf-8)
+    #+clisp (ext:convert-string-from-bytes octets charset:utf-8)
+    #+sbcl (handler-bind
+               ((sb-impl::octet-decoding-error #'(lambda (c)
+                                                   (declare (ignore c))
+                                                   (invoke-restart 'use-value "?"))))

Ben Spencer <dangerous.ben <at> gmail.com> writes:

> ---
>  wrappers.lisp |   26 ++++++++++++--------------
>  1 files changed, 12 insertions(+), 14 deletions(-)

Nice. Please commit. Works fine.

Regards,
--

-- 
Julian Stecklina

The day Microsoft makes something that doesn't suck is probably the day
they start making vacuum cleaners - Ernst Jan Plugge

_______________________________________________
Stumpwm-devel mailing list
Stumpwm-devel <at> nongnu.org
http://lists.nongnu.org/mailman/listinfo/stumpwm-devel

Hi,

On Tue, Jan 19, 2010 at 10:47:57PM -0500, Brit Butler wrote:
> I have a Lenovo X200 laptop with a dock, the dock is connected to an
> external flatscreen via VGA. With Xmonad I would get home, drop on the dock,
> use lxrandr to enable the second screen and be off to the races. This didn't
> work with stumpwm so I wrote the fix-layout defcommand in my .stumpwmrc
> which works handily.

What exactly does it not do that you expect it to?

> Connected to stump with slime, it's interesting to note
> that (current-screen) always returns the screen stumpwm first loaded with,
> even if I turn OFF the VGA monitor in lxrandr and turn on the laptop screen
> (or vice versa). (current-head) always shows the accurate resolution of the
> active screen. It's also interesting to note that *screen-list* never has
> more than one entry, including with both screens on in lxrandr.

Terminology note: 'screen' refers to X screens, 'head' refers to
Xinerama heads.  It sounds like you have two heads within one screen,
so I'd expect *screen-list* to only show one entry.  However, when you
have both heads enabled, (screen-heads (current-screen)) should have
two entries.  The command:

xdpyinfo -ext XINERAMA | grep head

should show dimensions that match these entries.

> So, the weird bug:
> When I "flip" monitors in lxrandr and hit OK, stumpwm hangs. The window I'm
> in receives all keyboard input and, generally, it's like stumpwm isn't there
> at all. If I switch to a tty, open emacs, slime-connect to stump and try to
> run (loadrc) or most anything in the :stumpwm package, I get a backtrace
> like that seen in http://redlinernotes.com/docs/stumpwm_backtrace.txt. The
> brief absence of the former (current-head) causes something not to
> type-check as head and even though a new head takes its place, stump hangs.
> Aborting to SLIME's toplevel in the debugger returns stump to it's normal
> state. I'm happy to try and produce fuller debugging information.

Are you running the latest from git?  This sounds rather like a bug
that was fixed relatively recently and isn't in the latest release.  I
get a 404 from that backtrace link btw.

> It would be nice in any case if anyone could tell me how to add a
> *head-changed-hook*, so that I can have fix-layout run automatically when I
> switch heads.

xrandr changes are handled in the :configure-notify event handler in events.lisp.

Hope some of this is helpful :)

Ben

On Thu, Jan 21, 2010 at 12:19:02AM -0500, Brit Butler wrote:
> I would expect the windows to expand to fill the (added) screen. The laptop
> runs at 1280x800, the VGA monitor at 1680x1050. If the laptop monitor is on
> and I connect and turn on the desktop monitor with lxrandr, the full screen
> real estate is not used. The laptop's 1280x800 resolution is simply mirrored
> on the larger screen. I have not found a simpler way to address this
> (getting *any window* at 1680x1050 resolution) than turning off the laptop
> monitor in lxrandr and iterating through all windows in all groups calling
> (redisplay). That is implemented as fix-layout in my .stumpwmrc and the
> laptop monitor must be *off* in lxrandr for fix-layout to work.

Are you aiming for a mirrored display or two separate heads?

Could you post the output of xrandr -q before and after switching on
the desktop monitor? (just trying to understand what lxrandr is up to).

> *version* returns "0.9.6-39-g545fd7a Compiled On Wed Jan 13 19:12:01". I
> built it from the stumpwm-git package on aur:
> http://aur.archlinux.org/packages.php?ID=12996
> In other words, I'm pretty sure I am...but I could always recompile. ;)

Yep, that's the latest.

> http://redlinernotes.com/docs/stump_backtrace.txt

It looks like you have *top-level-error-action* set to :break (some
instructions for setting up slime recommend this), which is why things
are locking up.  Setting it to :abort should hopefully give you a more
useful backtrace which you can grab with copy-unhandled-error.

(Side note: I'm not really sure what the practical difference between
:abort and :message is.  :abort throws out of the main loop whereas
:message just continues; so I guess the idea is that :abort ensures
you're in a stable state at the cost of a soft reset, whereas :message
leaves everything as it is, risking something somewhere being broken).

As to the bug itself, it looks like it's similar to the issue we had
before in that frame-head can return nil under certain circumstances,
and the caller doesn't expect it.

Ben

_______________________________________________
Stumpwm-devel mailing list
Stumpwm-devel <at> nongnu.org
http://lists.nongnu.org/mailman/listinfo/stumpwm-devel

> (Side note: I'm not really sure what the practical difference between
> :abort and :message is.  :abort throws out of the main loop whereas
> :message just continues; so I guess the idea is that :abort ensures
> you're in a stable state at the cost of a soft reset, whereas :message
> leaves everything as it is, risking something somewhere being broken).

I think :message may be old crust from how errors used to be handled.
It's possible one could simply want to display a message and push
through but I suspect that's almost always a bad idea.

-Shawn

On Thu, Jan 21, 2010 at 12:40:41PM -0500, Brit Butler wrote:
> Mirrored, with all windows in all groups taking advantage of the larger
> resolution. To be clear, I actually shut the laptop lid on the dock as there
> isn't enough room on the desk for both screens to be open and visible.

What happens if you actually disable the laptop screen?  Or does that
trigger the crash?

(using xrandr that would be 'xrandr --output LVDS1 --off', not sure
how you'd do the same thing with lxrandr)

Could you also post before and after output for:

xdpyinfo -ext XINERAMA | grep head

and

(screen-heads (current-screen))

the latter from slime.

Thanks, Ben

---
 contrib/mpd.lisp |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/contrib/mpd.lisp b/contrib/mpd.lisp
index 01a6e91..3bbb8ac 100644
--- a/contrib/mpd.lisp
+++ b/contrib/mpd.lisp
 <at>  <at>  -114,6 +114,7  <at>  <at> 
   #(127 0 0 1)
   )
 (defparameter *mpd-port* 6600)
+(defparameter *mpd-password* nil)

 (defvar *mpd-timeout* 50)

 <at>  <at>  -204,7 +205,9  <at>  <at> 
     (when *mpd-timeout*
       (setf *mpd-timer*
             (run-with-timer *mpd-timeout* *mpd-timeout* 'mpd-ping)))
-    (read-line *mpd-socket*)))
+    (read-line *mpd-socket*)
+    (when *mpd-password*
+      (mpd-format-command "password \"~a\"" *mpd-password*))))

 (defun mpd-ping ()
   (mpd-send-command "ping"))
--

-- 
1.6.4

5	April 2013
8	March 2013
18	February 2013
6	January 2013
7	December 2012
10	November 2012
101	October 2012
18	September 2012
19	August 2012
12	July 2012
14	June 2012
29	May 2012
17	April 2012
40	March 2012
17	February 2012
23	January 2012
25	December 2011
64	November 2011
26	October 2011
23	September 2011
16	August 2011
46	July 2011
34	June 2011
34	May 2011
15	April 2011
108	March 2011
14	February 2011
2	January 2011
12	December 2010
17	November 2010
35	October 2010
25	September 2010
14	August 2010
24	July 2010
28	June 2010
51	May 2010
24	April 2010
20	March 2010
8	February 2010
18	January 2010
46	December 2009
113	November 2009
37	October 2009
95	September 2009
34	August 2009
52	July 2009
53	June 2009
47	May 2009
89	April 2009
61	March 2009
29	February 2009
42	January 2009
40	December 2008
33	November 2008
131	October 2008
52	September 2008
98	August 2008
138	July 2008
75	June 2008
143	May 2008
30	April 2008
39	March 2008
79	February 2008
38	January 2008
14	December 2007
5	November 2007
55	October 2007
10	September 2007
19	August 2007
32	July 2007
5	June 2007
19	May 2007
8	April 2007
1	March 2007
20	February 2007
20	January 2007
7	December 2006
29	November 2006
18	October 2006
11	September 2006
4	August 2006
8	June 2006
8	May 2006
34	April 2006
36	March 2006
1	February 2006
4	January 2006
1	December 2005
1	September 2005
8	July 2005
2	December 2004
14	November 2004
7	October 2004
1	September 2004
11	July 2004
11	June 2004
4	May 2004
2	April 2004
8	March 2004
5	February 2004
2	January 2004
14	October 2003
13	September 2003
1	February 2002

[STUMP] Re: [PATCH] Modify pathname-is-executable-p to work with symlinks in recent SBCLs.

[STUMP] [PATCH] UTF-8 support for CCL.

Re: [STUMP] [PATCH] UTF-8 support for CCL.

[STUMP] Interesting Bug...

Re: [STUMP] Interesting Bug...

Re: [STUMP] Interesting Bug...

Re: [STUMP] Interesting Bug...

Re: [STUMP] Interesting Bug...

Re: [STUMP] Interesting Bug...

[STUMP] [PATCH 14/14] Added password identification support for mpd contrib.

Mon	Tue	Wed	Thu	Fri	Sat	Sun

				1	2	3
4	5	6	7	8	9	10
11	12	13	14	15	16	17
18	19	20	21	22	23	24
25	26	27	28	29	30	31