RK>Hate to burst your bubble guy but I can produce a 128 byte program that
RK>will make any HPFS partition not only unreadable but also unrecoverable.
RK>All one need do is punch a hole into the disk, that hole is nothing more
RK>than 1 cluster having it's format removed. One bad sector in the file
RK>system brings HPFS to its knees and that is quite simple to produce in
RK>very little code.  As for the lack of utilities, stability of the

Then we can assume that you did the right thing and forwarded said
program to Doug Azzarito or Mary Monahan at Boca Raton for analysis.
There are still some problems in Chkdsk under certain senarios which
could affect HPFS recovery. Doug has a standing offer to trace any
potential bugs that anyone has claimed to have found.  The CPS extent
not found problem blamed on HPFS 18 months ago was really a chkdsk
problem.  The HPFS file directory unlike the FAT file system can
be rebuilt from each individual files FNODE.  Therefore if there is a
bug it is in the recovery routine not the file system. Possibly Chkdsk
ate itself processing the hotfix map.  You say that this happened last
year.  Chkdsk has been reworked quite a bit since then.

I would also guess that your 128 byte program is run from native DOS
and not OS2 since it is doubtful that OS2 would let your program do a
direct access to the disk and I know of no 128 byte disk device
drivers.  I do not see any difference here that removing the master
boot record or blowing away the same section of both copies of the FAT.
Many DOS utilities run under native DOS are bigger viruses to an OS2
file system than anything you would come across accidently.  So the
point here is what's your point.

--Lynn

 * SLMR 2.1a * A problem can be found for almost every solution.

--- DB 1.39/004485