On Jul 11, 1996 at 08:58, Bob Lawrence of 3:711/934.12 wrote:

 DB>> Actually, I was referring to the fact that you said the relevant
 DB>> information was gone forever - Paul's just proved that it hadn't gone,
 DB>> was just .. "misplaced". :-)
 BL>
 BL> I said nothing about relevant information "gone forever."

Sure?  I was certain you told Paul to give up trying to restore his dates
and times because his stuffing around had consigned them to the great bit
bucket in the sky.  No doubt Paul (archiver of the ages) has the necessary
evidence to implicate the true culprit.

 BL> It uses two different tables, doesn't it, in different locations?

*Very* simplistically, HPFS splits the disk into 8Mb "bands" and
places directory information for each band at alternating start/end
locations, something like this:

[administrivia] [dir1] [band1] [band2] [dir2] [dir3] [band3] [band4] ...

Or something like that, anyway.

 BL> It has always struck me as totally *stupid* for DOS to put the two
 BL> FATs in adjacent sectors, and to keep only one boot record.

The FAT file system is an abomination - in theory, you can have any number
of positive integer FATs from one upwards .. but most software is
hard-coded for two, and expects them one after the other.  Of course, by
the time one is corrupted the data is copied straight over the second,
corrupting it as well.

The fact that Paul was able to recover so much is testimony to just how
reliable HPFS really is when it comes to the crunch.  With FAT, he would
have had no chance.

Cheers..

    - dave
    d.begley{at}ieee.org

---