Codepages be damned

Re. yesterday's blog, ok, so this is worse than I thought. Or perhaps better, depending on the way you look at it.

I broke down and booted into Windoze to try to figure out what codepage to use on the mount command, and lo and behold, regedit claims Windoze to be using CP 437, which is what I'd configured as the default in the kernel. But mounting from kernel 2.6.8 still doesn't work either specifying no codepage on the command, or cp 437.

Possibility 1 appears to be charset, which is another parameter on the mount command line.

Possibility 2 appears to be that Windoze is lying about what codeset the fs is formatted to.

Possibility 3 appears to be the new code is merely broken. I think I'd prefer this one, because while I'd owe some apologies re my uncharatable thoughts and comments, it is at least a reasonable mistake.

I think my next step is simply to back out the patch and produce a kernel-2.6.8--. Or perhaps check later kernels to see if someone further tweaked this area (but then those kernels might not have the various debian changes already incorporated into this one).

But coming back to "what is a user to do", I've gotta say this whole thing has been like a guessing game in that there isn't an apparent way for me to derive the parameter I need, and instead I'm left to guess what is expected. Ideally the mount error message would say "No, the drive is formatted to cp666 you moron! Use that!"

You: Pick a number between one and 100.
Me: Um, 42?
You: No! Wrong! Screwed! ....... Guess again!
... ad nauseum.

And I'm nauseated.