Re: [RP] [RFC v3] Add xrandr support

[Jeremie Courreges-Anglas]

Hi Rob,

first, thanks for the report.

Rob Paisley <address@hidden> writes:

> All:
>
> I run three monitors, two vertical and 1 horizontal.
>
> My login manager mirrors the two vertical monitors and ignores the horizontal
> one.
>
> Upon login, I open a shell in ratpoison (1.4.6),

While there shouldn't be relevant xinerama/xrandr changes between 1.4.8
and 1.4.6, it would really help if you used the most recent release in
tests.

> turn my head sideways
> (improper rotation of monitors) and type "sh bin/rotate.sh && exit
>
> rotate.sh is as follows:
> xrandr  --output DisplayPort-0          --auto --rotate left    --primary \
>         --output DVI-0                  --auto --rotate left    --right-of    
>   DisplayPort-0 \
>         --output DisplayPort-1-1        --auto --rotate normal  --left-of     
>   DisplayPort-0
>
> Then run ratpoison's newwm with ratpoison as the argument.

FWIW there's some syntactic sugar for this: "restart".

> This causes
> ratpoison to re-detect the screens and I continue my workflow normally
> with everything in harmony. ratpoison shows three screens, and I move
> between them with nextscreen/prevscreen as usual.

sigh, I didn't even know that such an xrandr-based setup was already
supported.

> This beta starts the same (as expected). The two vertical monitors are
> mirrored and incorrectly rotated. I expected that when I ran my xrandr command
> not to have to re-exec ratpoison to get directly to a correct setup. No
> change.
>
> I tried the newwm command and that seems to have decided that all three
> screens were one really big screen.
>
> Here's what I see from 'ratpoison -c sdump -c fdump' at various times.
> before: Right after login
> xrandr: After I've run xrandr from above
> after: After I've issued "newwm ratpoison"
>
> $ cat 1.4.6-before.txt 1.4.6-after.txt
> 0 0 0 1600 1200 1,1 0 0 1600 1200 0
> (frame :number 0 :x 0 :y 0 :width 1600 :height 1200 :screenw 1600 :screenh 
> 1200 :window 16777217 :last-access 0 :dedicated 0),
> 0 1600 0 1200 1600 1,1 2800 0 1200 1600 0,2 0 0 1600 1200 0
> (frame :number 0 :x 0 :y 0 :width 1200 :height 1600 :screenw 1200 :screenh 
> 1600 :window 16777217 :last-access 6 :dedicated 0),
>
> $ cat before.txt xrandr.txt reexec.txt
> 0 0 0 1600 1200 1,1 0 0 1600 1200 0
> (frame :number 0 :x 0 :y 0 :width 1600 :height 1200 :screenw 1600 :screenh 
> 1200 :window 18874369 :last-access 0 :dedicated 0),
> 0 0 0 1600 1200 1,1 0 0 1600 1200 0
> (frame :number 0 :x 0 :y 0 :width 1600 :height 1200 :screenw 1600 :screenh 
> 1200 :window 18874369 :last-access 5 :dedicated 0),
> 0 0 0 4000 1600 1
> (frame :number 0 :x 0 :y 0 :width 4000 :height 1600 :screenw 4000 :screenh 
> 1600 :window 18874369 :last-access 0 :dedicated 0)

With Mathieu's diff, here's what I get here

ritchie ~/src/ratpoison$ ratpoison -c sdump -c fdump
LVDS1 66 0 0 1280 800 1
(frame :number 0 :x 0 :y 0 :width 1280 :height 800 :screenw 1280 :screenh 800 
:window 14680077 :last-access 1 :dedicated 0)
ritchie ~/src/ratpoison$ ratpoison -c sfdump
(frame :number 0 :x 0 :y 0 :width 1280 :height 800 :screenw 1280 :screenh 800 
:window 14680077 :last-access 1 :dedicated 0) 66

Note how the output format for sdump is different.  I'm wondering
whether you actually have xrandr support built in.

I have pushed this xrandr feature on the "xrandr" branch on savannah,
with a couple more commits to tweak autotools and runtime xrandr
detection.

New tarballs:
  http://wxcvbn.org/~jca/distfiles/ratpoison-1.4.9-xrandr1.tar.gz
  http://wxcvbn.org/~jca/distfiles/ratpoison-1.4.9-xrandr1.tar.gz.sig
  http://wxcvbn.org/~jca/distfiles/ratpoison-1.4.9-xrandr1.tar.xz
  http://wxcvbn.org/~jca/distfiles/ratpoison-1.4.9-xrandr1.tar.xz.sig

Rob, could you please give this another try?  Please always specify
./configure --with-xrandr, and make sure you're actually running the
most recent executable.

Thanks,
-- 
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF  DDCC 0DFA 74AE 1524 E7EE