R devel - Jun 2011 - X11 device windows are spawned with empty WM_NAME property

Hello,

I've noticed that X11 device windows (the ones that pop up when I plot
something on Linux) aren't assigned any WM_NAME property until after
they've
been spawned.  This causes problems for users of tiling window managers, who
may wish to use the WM_NAME property to decide whether to float or tile the
window.

If I create a plot and then examine the window properties with xprop, I see
something like the following: WM_NAME(STRING) = "R Graphics: Device 2
(ACTIVE)".  However, by instructing my window manager (xmonad) to float all
windows with a matching name, I can tell that when the window was created
its WM_NAME property was empty.

Is this behavior intended?  Would be it be reasonable to instead create
windows with a generic name like "R Graphics" prior to updating them
with
the device number and active/inactive status?

I'm not familiar with the R code base, but I found that creating a plot
results in a call to the function in_do_X11 in the file
src/modules/X11/devX11.c.  In that function (around like 2869) there is code
to parse out the window title from the SEXP args parameter, and it always
seems to parse out an empty string.  If I manually assign title="R
Graphics"
at that point, I am able to successfully float R plot windows based on their
WM_NAME. (line numbers based on SVN revision 56131)

Is this actually an issue that needs fixing?  Should I create a bug report
for this?

Thanks,
Drew

	[[alternative HTML version deleted]]

On Jun 17, 2011, at 01:53 , Drew Frank wrote:
> Hello,
> 
> I've noticed that X11 device windows (the ones that pop up when I plot
> something on Linux) aren't assigned any WM_NAME property until after
they've
> been spawned.  This causes problems for users of tiling window managers,
who
> may wish to use the WM_NAME property to decide whether to float or tile the
> window.
> 
> If I create a plot and then examine the window properties with xprop, I see
> something like the following: WM_NAME(STRING) = "R Graphics: Device 2
> (ACTIVE)".  However, by instructing my window manager (xmonad) to
float all
> windows with a matching name, I can tell that when the window was created
> its WM_NAME property was empty.
> 
> Is this behavior intended?  Would be it be reasonable to instead create
> windows with a generic name like "R Graphics" prior to updating
them with
> the device number and active/inactive status?
> 
> I'm not familiar with the R code base, but I found that creating a plot
> results in a call to the function in_do_X11 in the file
> src/modules/X11/devX11.c.  In that function (around like 2869) there is
code
> to parse out the window title from the SEXP args parameter, and it always
> seems to parse out an empty string.  If I manually assign title="R
Graphics"
> at that point, I am able to successfully float R plot windows based on
their
> WM_NAME. (line numbers based on SVN revision 56131)
> 
> Is this actually an issue that needs fixing?  Should I create a bug report
> for this?
> 
A well-tested patch might be considered, but beware of the demons...

Notice that you can actually set the title via X11(title="Foo's
bar"). If you do that, it won't display the device number. So (a) you
wouldn't want to destroy that feature, and (b) how does the code that
inserts the "Device 2" bit know that it can do so? By checking for an
empty title? (Yup).
 
I notice that the initial title string could contain formatting characters, so
possibly, what you really want to do is to default it to "R Graphics:
Device %d"). Something of the sort is already happening with the quartz()
device.
> Thanks,
> Drew
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
-- 
Peter Dalgaard
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Email: pd.mes at cbs.dk  Priv: PDalgd at gmail.com

On Fri, Jun 17, 2011 at 1:57 PM, Drew Frank <ajfrank at ics.uci.edu>
wrote:
> On Fri, Jun 17, 2011 at 7:26 AM, peter dalgaard <pdalgd at gmail.com>
wrote:
>>
>> On Jun 17, 2011, at 15:32 , Prof Brian Ripley wrote:
>> >>>
>> >>> Is this behavior intended? ?Would be it be reasonable to
instead create
>> >>> windows with a generic name like "R Graphics"
prior to updating them with
>> >>> the device number and active/inactive status?
>> >
>> > Yes, no.
>> >
>>
>> Well, "yes, maybe", I'd say. If the WM insists on looking
at the title, it makes some pragmatic sense for the user to set it to a generic
name that the WM can recognize.
>>
>> However, on revisiting this, the obvious fix/workaround is just to set
>>
>> X11.options(title="R Graphics: Device %d")
>>
>> and the user can do this in his/her startup files. No need to inflict
it on every user with any WM.
>>
>> --
>> Peter Dalgaard
>> Center for Statistics, Copenhagen Business School
>> Solbjerg Plads 3, 2000 Frederiksberg, Denmark
>> Phone: (+45)38153501
>> Email: pd.mes at cbs.dk ?Priv: PDalgd at gmail.com
>>
Thank you Peter and Brian for your replies.? The X11.options
workaround is effective, and I'd be happy to make use of it.

That said, after reading your replies I agree that the "correct" thing
to do would be for the WM to match on WM_CLASS, which is currently not
set.? I checked out the R-devel trunk and was preparing a patch to set
WM_CLASS when I noticed that Brian already committed code to do this
in r56008. ?However, the code is wrapped in an "#ifndef USE_Xt", with
the comment "/* For those too idle to make use of Xt (PR#14588) */".
When I compile, USE_Xt is set so this code isn't executed and WM_CLASS
is never set. ?Is it supposed to be set somewhere else?

Thank you,
Drew