[R-SIG-Mac] problem with X11() after install of 2.7.0.

[Joerg van den Hoff]

On Wed, Apr 30, 2008 at 11:46:45AM -0400, Simon Urbanek wrote:
> 
> On Apr 30, 2008, at 11:31 AM, Joerg van den Hoff wrote:
> 
> >On Wed, Apr 30, 2008 at 10:55:38AM -0400, Simon Urbanek wrote:
> >>
> >>On Apr 30, 2008, at 10:34 AM, Joerg van den Hoff wrote:
> >>
> >>>hi everybody,
> >>>
> >>>today  I  decided  (unwisely,  it  appears) to install 2.7.0
> >>>under MacOS 10.4.7. now, when I  run  R  from  the  terminal
> >>>command line and do something like
> >>>
> >>>plot(1:10)
> >>>
> >>>a  X11 graphic window pops up, the 10 points are drawn on an
> >>>otherwise blank canvas (no axes, no labels, no  ticks  etc),
> >>>CPU  time  goes  to 100% and stays there and R does not come
> >>>back to the prompt ("hangs"). I tried to `ktrace'  this  but
> >>>this  produced a one Gig logfile in a few seconds from which
> >>>a learned nothing ...
> >>>
> >>>a quick search of the  archive  did  not  yield  a  hit.  so
> >>>please: what's going on here?
> >>>
> >>
> >
> >thanks a lot for this "real time" response.
> >
> >>Ancient system? As a very first step for your own good I'd recommend
> >
> >well, no. actually I _have_ 10.4.11 installed (I somehow was sure  
> >it's 10.4.7,
> >still. that's probably alzheimer casting his shadow :-)).
> >>updating your system which is way out of date (the current version
> >>Tiger is 10.4.11). Then if the issue persists, please report your  
> >>full
> >>sessionInfo() and machine used. The above works without problems on
> >>both 10.5.2 and 10.4.11 (Mac Pro and iMac).
> >10.5.2 is no option for me since I need a working fullscreen X11 mode.
> >I now checked it again both on a dual 2Ghz G5 PPC and a 1.5 GHz  
> >power book PPC G4,
> >both running
> >10.4.11: same problem. actually, R hangs only when really plotting. so
> >I can first open the device with X11() and get back to the R prompt.  
> >only
> >when executing the `plot(1:10)' R apparently hangs and everything is  
> >as described
> >in my first mail.
> >
> >here comes the session info:
> >
> >R version 2.7.0 (2008-04-22)
> >powerpc-apple-darwin8.10.1
> >
> 
> Hmm - I have tried even the PowerPC version now and it works without  
> problems for me.
> Are you using Apple's X11 or something else? Also do you get the error  
> in the "regular" (non-root) mode?
> 
> Thanks,
> Simon

simon,

as a follow up: I just stumbled over the  discussion on r-devel
concerning speed of the new X11 etc. devices. I understand cairo graphics
is used now which provides antialiasing but is slow (I'm sufficiently far
on the "want to get my work done with R" side that I'm not in sync with
what's going on in the area of graphic libraries development...).

anyway, in that mail there was a remark that using X11(type=Xlib) restores(?)
the old X11 device behaviour (fast, but "ugly" (I personally don't care for this
"on-screen")).

so: I just tried this with 2.7.0. on my power book: after starting R issue

X11(type="Xlib")
plot(1:10)

this just works as 

X11()
plot(1:10)

did with 2.6.2. that suffices more or less as a work around for me (but probably would imply a lot
of editing of existing code to open X11() everywhere correctly).

now comes the really strange point: since yesterday I've tried repeatedly to use

X11()
plot(1:10)

directly after starting 2.7.0 on both my PPCs with the result 
described in my previous mails to the list. but I now discovered that doing
instead

X11(type="Xlib")
plot(1:10)
graphics.off()
X11()
plot(1:10)

*does* produce the complete plot. and even more strangely, now closing R altogether, 
restarting and issuing directly

plot(1:10)

now works, too. I take every oath that I'm otherwise not seeing white mice
crawling all over the place. the first thing (X11() works after previously
opening X11(type = "Xlib") ) I might comprehend (something goes on in the
running R process) but the latter (now it works directly after restart)
I find really strange: what permanent changes to the file system/run time 
environment have occured?

what's going on here? the effect was definitely there. it now looks like
issuing _once_ X11(type = "Xlib") followed by graphics.off(); X11()
triggered some action which made the "new" X11() device working.
what could this be? 

all in all it seems my problem went away permanently (since now the
plots procede even after restart of R), but I'd say over all the
observed behaviour is sort of alarming. does somebody have any
idea what's happening here? 

regards,

joerg

PS: I just checked that evrything remains OK even after restarting
the X server.