Hi all, I am pleased to announce FORMAT 0.91s, which you can get at:
http://www.coli.uni-sb.de/~eric/stuff/soft/by-others/ format-0.91s.zip
(remove the space to get the actual URL)

Changes: Daniele Giacomini did several tests with his various old PCs
with 5.25in drives, and I think the (broken since 0.91q?) handling of
5.25in formats should now work better than in most older FORMAT versions  :-) .
FORMAT also avoids a false DMA overrun error (Mark Zbikowski wrote about it
in an old usenet posting, and somebody recently mentioned his other postings
on the freedos-devel list...) and saves/restores the DDPT between int 13
calls (otherwise, log printing, if redirected to another floppy, would mess
up the floppy property settings between the accesses to the to-be-formatted
floppy - found by Daniele, too). The DDPT log text is shorter now, and the
drive setup for floppy format got cleaned up. As suggested in Bugzilla entry
1823, FORMAT now checks the carry flag first before checking the returned
status (AH register) after int 13 (BIOS disk services) calls. This check was
missing in some floppy-related calls.

Bugzilla entries:
998 - lists many test scenarios, especially for FAT32. Currently open:
  FAT16 vs disk access (probably fixed now), inconsistency between DOS
  and boot sector for small FAT16 disks where DOS uses FAT12 for some
  reason, MIRROR might need a "confirm" prompt (I do not think so: The
  mirror step is aborted if it is found to be destructive, i.e. if the
  area where the mirror data will be stored is found to be in use), and
  the "type the full word yes" prompt is teasing the user  ;-) . Yves also
  mentions that the default format type for harddisk is quick (safe or
  irreversible) - fast but no surface scan / wiping. No problem IMHO.
  Yves updates his own report later - newer FORMAT solves the FAT12/16
  effect and disk access problems, but NDD/DiskMinder were not perfectly
  happy: They think small FAT16 partitions are no good idea (etc.)?
  And DiskMinder is SOMEHOW not happy with the FAT (16/32) for non-huge
  disks. I hope Yves can find more specific information here.
998 - in addition, there are still some formatting styles which have not
  been tested (at least not documented in bugzilla). Please have a look
  and report your experiences (Hi Daniele! ;-)).

1816 - format to 360k was broken, should probably be fixed in 0.91s  :-) .
1823 - carry flag should be checked after calling int 13, not only the return
  status in register AH. This should be fixed in 0.91s, too.


I ask the bug reporters to re-test with 0.91s and close bugs 1816 / 1823,
after adding a comment that 0.91s fixes the bug, if the test results confirm
that the new FORMAT actually solves the problems.

Enjoy the new FORMAT release 0.91s  :-) .

Eric

PS: Please let me know if you found that I introduced new bugs...! And let
me know if I forgot to fix old bugs. And let me know if 1816 / 1823 are fixed.
Thanks!