960121  Minor cleanup, released 2.5.4
951209	Added "magic" header to manpages to have man run them through tbl
951209	MTOOLS_LOWER_CASE is back, various small bug fixes over
	Tuesday's changes
951205	Bus strike in Grenoble! Well, let's do something useful and
	re-arrange the config file syntax :-)
	The syntax has become much more flexible now, and also
	includes items which used to be only accessible via
	environmental variables.
	Moreover, it is now possible to include character translation
	tables in line.
951126	Fixed another Atari disk bug: Atari disks sport a bogus
	"number of hidden sectors"
951125	Fixed missing zero-terminator in autorenamed long names
	MTOOLS_SKIP_CHECK now implies MTOOLS_FAT_COMPATIBILITY.
951124	Fixed small pine-induced typo in the Makefile.
	<rant on>
	Folks, please don't use pine. It sometimes changes the CONTENT
	of your messages. Case in point: 400 gets transformed into
	=A3400, which looks like 3400 pounds to a person unaware of
	this pine "feature".
	<rant off>
951123	Mformat now puts a 12 bit FAT on ED to better match Messy Dos'
	behaviour.
951115	Added ability to do mcopy e: to copy all files from the root
	directory of e:
	New Xdf-less Linux target in the Makefile
	Relaxed sanity check to let pass wonky Atari disks whose FAT
	begins with 3 zero bytes.
	Make the check of the initial fat bytes conditional on
	mtools_skip_check
	Corrected "testna=" bug
	Upped minimal sector size to be 256 (instead of 128). This
	helps 2m30
951112	2m30 compatibility
	Manpage update
	2m checksum bug fix
	Ability to mformat 2m disks
951107	Xdf bug fix (dev parameters always set to Xdf, even if it
	wasn't really an Xdf disk)
	Fixed YAHB (yet another hash table bug :) ) . Hope this one's
	the last.
	Centralising most env-var handling.
	Update of the mtools manpage.
	Xdf is now optional, and only active if MTOOLS_USE_XDF is
	set. Saves a few milliseconds of startup time on non Xdf
	disks.
	Some lawyer-proofing, just in case :)
951106	Fast xdf code (finally!)
	Minor performace enhancements here and there.
	Names which are all lower case now generate a long name entry
	(according to Steve Searle, that's how Win'95 behaves).
951029	Character translation table fixes. Other name fixes.
951026	Put restrictions on longnames to better match Win'95.
	(suggested by Steve Searle)
	Reworked autorename code. Catch SIGHUP signal
	Added missing file close to main loop
	Changed name of the "ask for action" commandline flag to 'm',
	and used 'a' for 'Autorename'.
951024	Removed infinite loop bug in hash.c, which occured when the
	hash table was filled with deleted entries.
951023	added Atari ST-style serial numbers (they live in the banner)
	fixed a troff bug in mtools.1
	Both changes were suggested by D. Hugh Redelmeier (hugh@mimosa.com)
950916	v2.5.3 released (after lots of fixes)
950904	v2.5.2 released
950904	mdir.c: initialized "files" and "blocks" to avoid complaint by GCC
	mattrib.c: initialized "code" to avoid complaint by GCC
950904	Based on commentm by Paul Slootman <paul@ahwau.ahold.nl>:
	init.c: fs_init(): initialized disk_size to 0.  The section
		which previously initialized this is ifdef'd out.  Why?
	Makefile, device.c: Passed ETCDIR (e.g. /etc or /etc/defaults)
		Gee, the Makefile is pretty ugly!  It might be good to start
		thinking about autoconfigure, or at least some cleanup.
	Makefile, mformat.c: use -DSOLARIS2, use srand48()/rand48()
	msdos.h, file.c, mdir.c: prefixed YEAR/MONTH/DAY/HOUR/MINUTE/
		SEC with DOS_ to avoid conflicts with <sys/time.h> on SVR4
	devices.c: use %i instead of %d to allow different bases
	parse.c: Changed comment for get_name()/get_path(); need to
		revisit this after deciding on default case behavior
	devices.c: load_devices(): fixed bad fprintf, line 748
	parse.c, mformat.c, misc.c: replaced expressions like
		  "if (islower(foo)) foo=toupper(foo)" with "foo=toupper(foo)"
		
950829 v2.5.1 released

950829 Based on comments by Martin Kraemer <Martin.Kraemer@mch.sni.de>:
		Bug fixes for compile errors and core dumps under SINIX-D 5.41
		(Siemens SVR4):
		plain_io.c
		mk_direntry.c
		vfat.h

950822 v2.5 released

950820 DCN	Change null-fill for unused remainder of VSE to 0xff fill for
		both upper and lower character (just one null for termination)
		This seems to better match Win95's behavior; Win95 had been
		complaining about bogus characters
		file_name.c: unicode_write()

950820 DCN	Commented out enforcement of VSEs being in order.  Win95 likes
		to put them exactly backwards, so we'd better tolerate getting
		them any way they might come!  Not sure what is lost by losing
		these checks, but it seems to be OK.

		directory.c: dir_read()


