Wow, that's rather large demonstration program. I managed to configure and build it. (But it hardcodes the path c:/msys/home/yoshi/... so it doesn't find any icons at run time, but hopefully that shouldn't matter for this alleged problem.) But I don't see any "blurry" panels. Could you provide a screenshot? And maybe a single source file demostration program, attached directly to this bug? Do you see this "blurriness" with gtk-demo? You should also try the latest GTK+ version from the maintained branch, 2.10.6.

Hello ,

This problem is just only on MS Windows OS. I Have tested my Alfa-Centauri program on Fedora Core 5 a Gentoo and it works (On Gentoo i have problems with gtkrc theme ,some parts of it are not ,,interpreted" properly ,with same gtkrc i get different 
look of application ,but i dont think that is problem of gtk+).I had also reported 
other bug which too only on Windows (resizing of headers in treeview)

Here are some screenshots:
http:/gtkuser.szm.sk/AlfaCentauri.tgz

I have used following files to build application

                        glib-2.12.1.zip               pango-dev-1.12.3.zip
atk-1.10.3.zip          glib-dev-2.12.1.zip           pkg-config-0.21.tar.gz
atk-dev-1.10.3.zip      gtk+-2.8.20.zip               zlib123-dll.zip
cairo-1.2.2.zip         gtk+-dev-2.8.20.zip           
cairo-dev-1.2.2.zip     libiconv-1.9.1.bin.woe32.zip  
gettext-0.14.5.tar.gz   libiconv-1.9.1.tar.gz         
gettext-0.14.5.zip      libpng-1.2.8-bin.zip          
gettext-dev-0.14.5.zip  libpng-1.2.8-lib.zip          
glib-2.12.1.tar.bz2     pango-1.12.3.zip  

            See ya!
                    Thanks for development of gtk+
                                                   Mark Kica

Why not simply put the screenshot images up as such on the web? You don't win anything by putting already compressed png images into a compressed tarball, it just makes it more work for me to fetch the tarball, unpack, and view the images.

But OK, I did that. I don't see any blur, I see screen update errors. I have noticed such occasionally myself, and I do see them in the copy of your demonstration program I built. Do I think it's a big deal? Not really. Will it be fixed? Eventually maybe. An absolutely minimal test program would be nice, to minimize the amount of irrelevant logging output when tracing and debugging this problem.

Is this a dup of 144269?  I don't think the image tarball is even there any more and don't really want to fetch it either...

We are seeing this same problem in Wireshark for Windows, but not Unix (both GTK 2.10.6).  We are tracking the issue as Wireshark bug number 958 (http://bugs.wireshark.org/bugzilla/show_bug.cgi?id=958).  I will attach a screen shot of what it should look like, what it looks like when the problem first develops and what it looks like after it has fully developed (see below for steps to reproduce).

Steps to reproduce using Wireshark or gtk-demo:
Wireshark - Drag one of the dividers between the three panes of data up and down a few times after capturing some traffic.  Dragging the mouse cursor across the area where the problem is seen makes it go away piece by piece.

Gtk-demo - Go into Text Widget->Multiple Views and drag that pane divider up and down.  Moving the divider up and down rapidly seems to make the problem worse.  Clicking on the text behind the artifacts makes it go away.

Created attachment 80097 [details]
Image of Wireshark before the problem occurs

Created attachment 80098 [details]
Image of Wireshark as the problem starts to occur

Created attachment 80099 [details]
Image of Wireshark after the problem has become its worst

This has been a problem for awhile, and it's a duplicate of bug #144269.  It would be helpful if you post your screenshots on that bug report.

*** This bug has been marked as a duplicate of 144269 ***