Hi, I have been running some tests with and without xvfb-run.
In short, they fail in xvfb-run, and succeed on my framebuffer. I am thinking that perhaps xvfb-run emits stderr chatter when used with GL applications. $ xvfb-run glxinfo glxinfo name of display: :99.0 Xlib: connection to ":0.0" refused by server Xlib: Client is not authorized to connect to Server ... I wonder where those Xlib errors come from. They seem to be spurious side-effects, as glxinfo continues to run in a reasonable manner. (No, I'm not asking for "direct rendering: Yes" in xvfb :-) I also see Standard Error: Xlib: connection to ":0.0" refused by server Xlib: No protocol specified as well. What brought this up were the 222 errors in http://www.vtk.org/Testing/Sites/makapuu/Debian-vtk_4.0-1_i386.deb/20020909-0522-Experimental/Test.html which was using xvfb-run, compared to http://www.vtk.org/Testing/Sites/makapuu/Debian-vtk_4.0-1_i386.deb/20020909-1101-Experimental/Test.html which did not use xvfb-run. ( For the big picture on VTK testing, see: http://www.vtk.org/Testing/Dashboard/20020909-0300-Nightly/Dashboard.html ) I'm looking for a way to avoid those test failures with xvfb-run, and I am not sure if it is easier to change the testing software to be more accepting of xvfb-run output, or make xvfb-run quieter. -Maitland
