I was dropped from the beta program when I declared that SCANDISK was a
disaster, and would eventually require a recall of DOS 6.00, When they did
the "FREE" "Stepup" from 6.00 to 6.20, they let it continue to be blamed
on the disk compression, although SCANDISK was actually the cause of the
specific problems that warranted the replacemnt. Details available on
request.
On Wed, 19 Nov 2025, Anders Nelson wrote:
Elaborate please!
OOPS!
MAJOR ERROR
I meant SMARTDSK, th disk cacheing program
Nothing [much] wrong with SCANDISK. Well, DOS 7.00 SCANDISK wouldn't do
"EX-FAT", nor reset the "dirty bit", rendering an entire drive useless and
"unrecoverable" by normal users.
(Monty Python mentioned "naughty bits", not "dirty bits")
Win 3.11 installed SMARTDSK at the beginning of install.
And, by default, it did WRITE cacheing (with any write it would declare
that the WRITE had been completed without error, before it even tried).
If there was a hard error, the only critical error option was RETRY. If
that didn't recover, the entire install had to be forcibly aborted, with
NONE of the successful part left on the disk!
When I told Microsoft Beta management about it, they said "that's a
hardware problem, not our concern"; They didn't appreciate my attempt to
explain that when the OS encountered a hardware error, and knew about it,
it had an obligation to gracefully rcover, work around, or exit, not jam
up.
SSTOR and SPINRITE could find no errors on that part of the disk! But
Install always broke at the same place. I copied a hadful of nonsense
space-taker files onto the disk, so that Install would be writing to a
slighly different place on the disk; that worked!
Without write cacheing, you could/would, after retry, write down which
file had failed, IGNORE the error, and go back later to manually
re-install that file. (one of the very few situations where IGNOREing an
error was justified)
If the user shut off the computer before the write actually got done, . . .
I had a girlfriend who went back to college; in the morning, she would
stand, with her coat on, pulling on the paper while her homework printed.
Then, the instant that the DOS prompt returned, she hit the power switch.
The write cacheing also re-arranged the sequence of writes "for greater
efficiency"; instead of writing the file, and then writing the directory
entry on each file to be written, it would write all of the directory
entries, and then write the file contents.
If the user shut off the computer before that completed, . . .
Besides user actions, a power failure would cause the same problems.
Yes, I know; REAL computers have uninterruptible power supplies?
Using a laptop, instead of a desktop, also mitigates most of the power
interruption problems.
For DOS 6.20 "To repair the problems with compression", they made WRITE
cacheing off by default; if the user deiberately turned write caching on,
it would not rearrange the write sequence, and, when a progam exited, it
would not give the DOS prompt until all of the writes were completed.
THAT fixed the worst of the "problems with compression", that people
griped about.
--
Grumpy Ol' Fred [email protected]
