R data editor not usable when desktop effects enabled

Bug #472408 reported by oudalrich on 2009-11-03
This bug affects 3 people
Affects Status Importance Assigned to Milestone
r-base (Ubuntu)

Bug Description

Binary package hint: r-base

When desktop effect are enabled, R's data editor is unusable. The window contents are garbled from the start, and when scrolling through the data using the arrow keys the contents of the editor window is not redrawn properly. I have attached a screenshot that shows an editor window after scrolling around a bit. I have verified that the problem exists on two different machines with different video hardware (Intel and ATI)

Steps to reproduce:
1. Make sure desktop effects are enabled.
2. Start R
3. Type:
View(data.frame(matrix(rnorm(100), ncol = 5)))
4. In the editor window that appears, use the arrow keys to scroll around. Notice how the contents of the window get increasingly garbled.

ProblemType: Bug
Architecture: i386
Date: Tue Nov 3 10:25:27 2009
DistroRelease: Ubuntu 9.10
ExecutablePath: /usr/lib/R/bin/exec/R
Package: r-base-core 2.9.2-3ubuntu1
 PATH=(custom, user)
ProcVersionSignature: Ubuntu 2.6.31-14.48-generic
SourcePackage: r-base
Uname: Linux 2.6.31-14-generic i686

oudalrich (uhkeller) wrote :
Logan Lewis (lkubuntu) wrote :

I can confirm this behavior on an x86_64 build, but an otherwise similar setup. I'm running the Ubuntu 9.10 version of the Intel graphics driver on a "GM965/GL960" controller.

I did try making an exception for the window decoration and shadows for the R Data Editor window, but the problem persisted.

ramon.m.c (ramarcos01) wrote :

Same behaviour on

2.6.31-16-generic #52-Ubuntu SMP Thu Dec 3 22:00:22 UTC 2009 i686 GNU/Linux
00:02.0 VGA compatible controller: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller (rev 0c)

Desktop effects enabled.

oudalrich (uhkeller) wrote :

I have posted an R bug report at
but according to the reply from Peter Dalgaard, this is an Ubuntu/Compiz specific problem. However, it's not totally Compiz-specific on my system since I also see (less severe) drawing problems on Metacity. If I View() a large matrix or data.frame, parts of it in the upper left corner are sometimes not drawn until I press a key or move the mouse. In some cases just the grid around the row names is affected, sometimes the row names themselves, sometimes even entries in the first columns. Steps to reproduce:

1. Make sure you're running Metacity.
2. Start R.
3. Run this in R and make sure not to move the mouse or press a button afterwards:
View(matrix(rnorm(1000000), ncol = 1000))
4. If the viewer window looks ok, close it and repeat from step 3.

I have attached another screenshot that shows the drawing problem in Metacity. Could somebody verify this? And if somebody has access to a non-Ubuntu Linux machine with R, could you check if the problem really is Ubuntu-specific?

ramon.m.c (ramarcos01) wrote :

Just another point in the discussion: I used to have KDE in the near past (Kubuntu 9.04) and I didn't have any problems with data editor in R... even with desktop effects enabled. Does this mean it is Compiz-specific problem??

Sorry for not being more specific, but I remember having this issue with previous Ubuntu versions (9.04, I guess).

ramon.m.c (ramarcos01) wrote :

I cannot reproduce the issue after deactivate compiz (metacity --replace).

Logan Lewis (lkubuntu) wrote :

I can't reproduce the drawing issue with metacity --replace either. I haven't yet tried it under KDE 4 with desktop effects enabled.

I did, however, try compiling the latest R from source. It required libxt-dev (which I got from the 9.10 repositories), but it had the same problem under compiz. I'm not terribly familiar with the underlying workings of the R code, but the xt library is pretty low-level for X11 programming. I googled briefly for bugs with compiz and libxt, but only turned up some old Java bugs. There's obviously still more work to be done to figure out if it's a general Compiz problem, something specific to the Ubuntu build of Compiz (or libxt/X11), or something bug within R which only affects this particular setup.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers