Mac OS X 10.4 bug with ColorSync display profiles and fast user switching: Different symptoms and a fix that no longer works?

Posted by Pierre Igot in: Macintosh
August 30th, 2006 • 9:07 am

It’s been nearly two years since I first submitted a report to Apple regarding the bug with fast user switching and calibrated display profiles in Mac OS X 10.4. (I only posted about it when 10.4 was officially released, but I reported it to Apple as soon as I encountered it in early seeds of the new system.)

Sadly, as far as I can tell, the bug still has not been fixed by Apple. In fact, I was even able to reproduce it on a brand new MacBook running Mac OS X 10.4.7 for Intel.

In fact, that experience proved to me that the bug doesn’t just affect people with dual-monitor set-ups and with calibrated display profiles. I saw it on a MacBook with no calibrated display profiles and without a secondary monitor attached to it.

Why Apple appears to be unable to reproduce the bug and fix it is beyond me, especially since it is a pretty annoying bug for those who use fast user switching on a regular basis. (Fortunately for me, I only have to use it intermittently on my G5 Quad. I use fast user switching much more often on our PowerBook G4, also running Mac OS X 10.4.7, but I have been unable to reproduce the bug on that machine.)

While doing some testing with another user environment on my G5 Quad the other day, I experienced the bug again upon returning to my regular user environment. The symptoms are that and the display calibration suddenly changes and everything becomes washed out, and the “Color” pane of the “Displays” preference pane becomes unusable, with system beeps all over the place.

Only this time I noticed that things did not become as washed out as they normally did with this bug. There was definitely a shift in the overall brightness of the picture when returning from the second user environment to the first user environment via fast user switching, but the shift was not as pronounced as it used to be.

As well, the only way to eliminate the problem entirely once it occurs is to restart the computer. But a while ago someone posted a fix at Mac OS X Hints that involves the use of a command called “DMProxy.” The command does not restore the functionality of the “Displays” preference pane, but at least it eliminates the brightness shift and goes back to your preferred display calibration.

However, when I tried this fix the other day after the more subtle brightness shift, I noticed that it no longer seemed to work. I tried it several times, and I could not get it to restore my preferred display calibration. I ended up having to restart the machine.

So to me it looks like something has changed, at least on my machine. I don’t know if it’s due to a recent system update or to some other troubleshooting procedure that I recently did for something else that was entirely unrelated but might have ended up having an impact on that bug.

The other troubleshooting procedure is a long story. In a nutshell, after much head scratching about a software bug (a keyboard shortcut that wouldn’t work) in a third-party application that the developer himself was unable to reproduce, I used a procedure that consisted of removing all my preference files from my home library’s “Preferences” folder, logging out, logging back in, and then putting all the preference files back in (replacing the new ones created by Mac OS X in the process with the old ones), logging out again, and logging back in.

It sounds like this should not help with anything in anyway, since you are effectively putting back all the preference files back where they were. But believe it or not, that particular procedure managed to eliminate the software bug in the third-party application.

This too me indicates that there can be some rather mysterious things going on behind the scenes in Mac OS X. And the bug with fast user switching certainly appears to belong to the same category, which is why I suspect that the unrelated troubleshooting procedure described above might have had an impact on the specific symptoms of this bug and might explain the different symptoms and the fact that the DMProxy fix no longer works.

Unfortunately, that’s about I can say or do about this particular problem. If someone else out there has the bug with fast user switching and the color shift, and is willing to try the troubleshooting procedure described above and see if it makes any difference, let me know :).

2 Responses to “Mac OS X 10.4 bug with ColorSync display profiles and fast user switching: Different symptoms and a fix that no longer works?”

  1. matsw says:

    I think your procedure does not work the way you intend. Say you overwrite the pref file of component A, called with some older version, while logged in. When you log out, component A will write its prefs from memory to disk, overwriting the older version you put there. The only way of making sure you get the right version is to manipulate the preference files while logged out (e.g. from another account or using shell commands and ssh).

  2. Pierre Igot says:

    I have always wondered about this. Obviously some parts of the system modify their prefs file only when you quit them, whereas others modify them as soon as you make a change.

    It is true that my “procedure” probably leads to some older pref files being overwritten by new files when quitting and relaunching the system. For example, after that procedure, I still had to reset some things in the Finder prefs and in the Appearance pref pane.

    All I can say is that that procedure did indeed cure the keyboard shortcut conflict that I had in that third-party application, and that it appears to have changed the behaviour of the fast user switching bug somewhat.

    So possibly both the keyboard shortcut and the fast user switching bug involve one of the pref files that was overwritten by a new file even after I had restored the older files.

    The question, of course, is: Which one? That is still a mystery to me.

    Thanks for your comments.

Leave a Reply

Comments are closed.