############################################################
List of new features in Mule Ver.2.3     -- 95.7.24

1. This version is still based on GNU Emacs 18.28, and
changed are mainly for bug fixes except for the following
topic.

2. By the cooperation with Omron Software, we can include a
feature to use jserver of Wnn6 as a kana-kanji conversion
server.  See INSTALL for more detail.  General information
about Wnn6 is avairable at:
	http://www-sp.omronsoft.co.jp/

############################################################
List of new features in Mule Ver.2.2     -- 94.12.28

1. This version is based on GNU Emacs 19.28 and changes are
mainly for bug fixes.

2. `m2ps' is updated and can now handle any of text coded in
*internal*.

3. New command `coco' is available for code conversion among
any of coding-systems supported by Mule.  `coco' package
announced in the previous version is obsolete.


############################################################
List of new features in Mule Ver.2.1     -- 94.10.29

1. The way to specify fontset gets easy.  See mule/fonts
section of info for more detail.

2. Improvement of Quail

2.1 Works also in minibuffer.  The guidance is shown in a
window which pops up at the bottom of screen.

2.2 Quail packages can be compiled.  A compiled package
is load very quickly.

2.3 Cxterm's tit format to quail-package converter is
provided in lisp/quail/tit.el.  All quail packages for
chinese are updated.

3. Arabic support improved.

4. Ethiopic support starts.

5. Direct input of Big5.  Now `set-keyboard-coding-system'
accepts '*big5*.

6. M-x mule-diag shows you useful diagnosis about the
current state of your running mule.

7. Multilingual Xmenu support.

8. cocolib package is distributed under `contrib' of Mule's ftp directory

This package contains a generic code conversion program.  In
addition, the program `any2ps' in this package is a
convenient front-end of `m2ps'.  Since `any2ps' accepts any
coding-system, you don't have to save text in *internal*
every time before using m2ps.


############################################################
List of new features in Mule Ver.2.0     -- 94.8.6

1. Now Mule is based on GNU Emacs 19 (.25)

Of course, Mule supports GNU Emacs 19's various new
features.

2. You can use `face' of Emacs19 even on terminals.

By `configure'ing Mule with the argument `--terminal-face',
such faces as underline and highlight get effective even
when you run Mule on terminal (i.e. -nw).

3. Thai character-set gets official

Since TIS620 has been registered in ISO, Mule also handles
it as an official character set.  The new leading-char for
TIS620 (lc-thai) is 0x85.  If you wrote Thai text with
coding-system *junet*, the escape sequence used for Thai
characters is not the same as that of previous version of
Mule.  But, you can still read a Thai text file of old
escape sequence.

4. Experimental support of Arabic

When you load arabic.el, you can write Arabic text.

5. Stop supporting BIG5-HKU

Since HKU version of BIG5 is just an error version, Mule
stops supporting it.

6. CCL is completely changed

To improve speed and easiness of writing, CCL syntax is
changed completely.  Read `CCL (mule)' node in Info for more 
detail.


############################################################
List of new features in Mule Ver.1.1     -- 94.2.10

1. CRLF is now handled automatically.

By the big change in the structure of coding-system, we now
can automatically detect how end-of-line of a text is
encoded and choose an appropriate coding-system for a new
buffer.  Please read doc/coding.txt for more information.


############################################################
List of new features in Mule Ver.0.9.8   -- 93.6.20

1. CRLF is not handled automatically.

The facility of automatic detection of EOL type (LF, or
CRLF) is deleteted because:
	o causes hard problem in process I/O
	o not necessary if code-priority is defined appropriately
Instead, on defining coding-system, you can specify the EOL
type as one of LF (Unix), CRLF (DOS), CR (Mac).

2. Vietnamese support

Now you can read/write files encoded in VISCII, VIQR, VSCII.
Vietnamese inputting method is also supported by QUAIL.

3. More rich support for right-to-left writing language

For right-to-left characters, we have added two features:
	o r2l-double-cursor (Emacslisp variable)
		When cursor is in between right-to-left char
		and left-to-right (normal) char, you can see
		insertion points for both types of character
		by two cursors.
	o highlight-reverse-direction (Emacslisp variable)
		A character whose direction is different
		from that of the current buffer is
		highlighted by gray-color (or background mesh).
Both mode is on by default.  Try these while you are
writing, for instance, Hebrew.

In addition, ISO6429's escape sequence for specifying
directionality is used on I/O.

3. CCL -- Code Conversion Language

Now Mule has a interpreter of CCL (simple stack language for
code-conversion) for non-ISO2022 coding-system.  It's much
faster than doing the same thing by Emacslisp.  CCL can also
be used to map a character code to that of a font while
displaying on X.

4. Quail packages for Chinese get more combenient

Now it works much similar as cxterm's inputting methods with
the features of showing candidates dynamically and selecting
candidate by ".," and digits.


############################################################
List of new features in Mule Ver.0.9.7.1   -- 93.1.22

1. Usage of EGG (TAKANA)

Although EGG can communicate with both jserver and cserver,
using EGG for both Japanese and Chinese was not convenient
in the previous version.  Now the environment is improved.
But please note that the previous command set-wnn-host-name
is changed to set-jserver-host-name and
set-cserver-host-name.

2. Thai inputting methods

Quail-package of the name 'thai' is ninproved.  You can
enter a composite character easily.  For instance, in
quail-mode for Thai, typing CVT (where C:consonant
V:upper-vowel, T:tone-mark) will produce a character
something like:

	  T
	  V
	__C__  <-- base line

Though, we still have a problem in deleting only T or V.

3. Automatic handling of CRLF at end of line

Mule's automatic coding-system detection mechanism can now
detect the type of end of line.  In the world of DOS, the
end-of-line mark is the sequence of CR and LF.  When mule
reads such a file, it detects the sequence and set an
appropriate coding-system for the file.


############################################################
List of new features in Mule Ver.0.9.7   -- 92.12.26

1. Composite character

You can compose any characters (upto 16 chars) into one
composite character, in that case, all the characters are
displayed on the same column and treated as one character.

But!!, for the moment, vertical split of window is not
supported for composite character.  If you split window
vartically while showing composite characters, I don't no
what happens. :-(

'm2ps' is also updated to support composite character.


2. Thai character support

By utilizing composite character, now you can treat Thai
character on X window.  See doc/demo for example.
lisp/quail-thai.el provides input method of Thai, the
program is not yet completed.  We are waiting a contribution
by who has knowledge of Thai writing.


3. Right-to-left displaying

Each character set has new attribute 'display-direction'.
Although, most of character sets have 'left-to-right'
feature, Hebrew and Arabic character sets have
'right-to-left' feature.  Characters who have this feature
are displayed from right to left on the screen.  See
doc/demo and try to move cursor around Hebrew letters.

But!!! (again), the support is yet quite buggy.  You may
find many many inconveniences.


4. Coding-system is now a symbol

You may not find any change on user-interface, but emacslisp
programers will find that handling coding-system gets much
easier.


############################################################
List of new features in Mule Ver.0.9.6   -- 92.10.26

1. EGG requires Wnn4.108.

You need Wnn4.108 to compile Mule with '#define EGG'.


2. TAKANA of EGG provides Chinese inputting facility.

Now, TAKANA can communicate with cserver.  All setups for it
is done in lisp/chinese.el.  To use this facility, you also
need SiSheng font (available in fonts/ETL.tar.Z or in
fonts/SISHENG.tar.Z).  The default setting uses the font in
ETL.tar.Z.


3. QUAIL system updated.

You must type 'C-]' (not 'C-\') to startup quail.
'quail-latin.el' supports Latin-2,3,4,5 inputting.


4. Support for new machines and systems

NeXT, PFU-A-Family, 386BSD, Linux, VMS, and MS-DOS are
supported.


5. Required PURESIZE is reduced a little.

By autoloading some libraries.