Although it might workout... i don't think that the basic principle of running an update on an OS that won't boot in the first place, in order to “fix the problem†is entirely sound. More likely than not, the the two problems are related. I.e., though the search for some damaged "version" file to make Installer happy is possibly valid... chances are that whatever is preventing the current OS from booting is also what's actually causing the update process to fail at such an early stage.
I have to repair damaged operating systems on the average 3 times per week, and I have determined that first trying the combo updater and then trying an archive and install is the fastest, most effective, least disruptive route to go. About 85% of OS damage I am able to completely fix with a combo updater. 13% requires an archive and install. The remaining 2% requires much more work such as moving data, fresh install, and manually reimporting user accounts.
Although A&I would fix anything a combo updater fixes, it requires a lot more time to run, and uninstalls some drivers. (and I'd have to run the combo updater after it anyway) So I try to avoid that if possible. I'm always mindful of performing the least invasive/disruptive repairs practically possible.
What is the version number of the current (unbootable) OS?
Does it boot into SU mode?
Does it boot into Safe Mode?
I assume it has been DiskWarrior-ed to death, and that all appears normal from that angle. [yes?]
The quick fixes didn't help and I've already determined beyond reasonable doubt that the OS must be repaired with at least a combo updater.
Is doing a fresh 10.6.x install impossible for some reason, or merely inconvenient?
See above.
Is it possible that the original "Leopard-->Snow Leopard" upgrade history for this Mac was one that may have been limited in some manner (where the 10.6 disk was a special deal which was linked to the machine and its original hardware somehow) ?
I have seen where previous issues can cause future updates not to run. It doesn't really matter what caused it, the damage is done, be it from the IO errors, (most likely) or some other prior cause. I've ran into this problem on several occasions in the past, that either required me to A&I or reinstall and restore d data, and I don't like how disruptive a process this is for the user and would like to take the less disruptive combo updater option.
Also, I have several times encountered a similar case where the OS installer would not start for the exact same reason. So this problem can sometimes force me to do a clean reinstall, which is one of the most disruptive repairs I ever have to make. And it's sad to see the PC tech with windows, that is one of the most common required fixes since registry damage is common and usually forces a reinstall. If
nothing else I'd like to say I can fix it better than
windows In this case I got lucky in that the 10.6.3 (A) allowed itself to run without the same error, and (B) retained the users and settings. Frequently when forced to A&I under these conditions it basically refuses to import directory services, nuking accounts but importing home folders, requiring resetup. So this one was nicer than it could have been. I suspect there is a link between DS damage and this entire problem. I am still searching for answers as to what causes the root error message in the installer, and also where it determines installed OS version. I have seen cases where the software KNOWS what version of OS is installed even though the systemversion.plist is GONE, I get the impression that one file is for 3rd party use and apple relies primarily on a different file - I need to know where that is.