--------------------------
$Header: f:/itrans/rcs/changes 1.18 95/02/22 21:32:01 avinash Exp $
--------------------------
****** Version 4.02, 15 Dec 1994
****** Version 4.03,  1 Jan 1995
****** Version 4.04, 21 Feb 1995

    This version makes changes that are incompatible with 3.xx or earlier
    ITRANS texts. Markers have changed, but no transliteration codes have
    changed -- so old ITRANS encoded texts will be accepted as is.
    
    1) Support for CSX (Classical Sanskrit Encoding) texts.
       See doc/icsx.itx (or better still, just print out doc/icsx.ps).
    2) All ITRANS commands now begin with # (previously, \ was used, but
       that confuses ITRANS commands with TeX commands).
       Short markers have been changed to ##
       Only one short marker is available, and it acts as a toggle.
       To convert pre-4.0 docs to 4.0, a sed script is provided.
       If you do not have sed on your machine, you will have to manually
       make the modifications to your input files.
    3) Added TeX macro file (lib/idevn.tex) for easier use of Devanagari fonts.
    4) Fixed the hindi ra-faar bug. RA-<cons>-<cons>..<vowel> would not
       be handled correctly in previous releases. So, rdhva, etc input
       printed incorrectly. Fixed this problem, so now rdhva, etc work ok.
    5) Added a sample input file - doc/sample.itx. Beginning (or advanced)
       users should take a look at this small file.
    6) "M" now works like anusvara (".n").

    Ver 4.01:
    1) Added m-overdot (== anusvara) to CSX code.
    2) Fixed bad bug in idevn.tex - \par redefinition in \obeyspaceslines
       would eat up any numerical text at the beginning of a line!
       (Needed a \relax at the end.)
       
    Ver 4.02:
    1) Added | as a command in idevn.tex to print Da.nDaa when using
       Frans's font.
    2) Modified icsx.itx to make it a bit clearer that uppercase characters
       have different meanings than lowercase chars, even in CS/CSX mode.
    3) Added the file csx2i.c to the src/ directory, is a standalone
       filter program to accept CS/CSX input text through stdin, and produce
       ITRANS encoded text on stdout.
       
    Ver 4.03:
    1) Bug fixed: "tryaa" now prints correctly (was printed
       as ta-half, yaa, ra-faar! Should be: ta-ra conjunct, yaa).

    Ver 4.04:
    1) Removed the "aM" ilex.l rule, since "M" itself is recognized as
       anusvara, and so there is no need for "aM" rule. This does not
       matter for devanagari, but for other languages, which do not have
       aM rules in the .ifm files, but do have anusvara, this works out
       much better.
    2) TeX output now suppress { } characters, and prints them as char codes
       in the .tex output files, since { and } are TeX characters used
       for defining scope and having ITRANS print out { } makes it very
       confusing to search for matching braces in .tex files.
    3) DOS executable now made with the latest flex (2.4.7), supposed
       to fix some 8-bit input bugs.
    4) Added sample makefiles for Linux (gcc/flex), Sun, SGI, etc.
    
****** Version 3.00, 21 June 1992
****** Version 3.01, 17 July 1992
****** Version 3.02, 7 August 1992
****** Version 3.10, 11 September 1992
****** Version 3.11, 5 October 1992
****** Version 3.12, 15 October 1992
****** Version 3.20, 10 December 1992
    1) Changed Ri, RI, Li, and LI to R^i, R^I, L^i, and L^I, respectively.
       These vowels are not used much anyway, and I needed "R" for
       a consonant (telugu, tamil, and bengali).
       Changed chandra-bindu to .N (was .C)
       Changed "ng" to "N^", changed "jn" to "JN". These are minimally
       used consonants, but caused problems for na-ga, and ja-na ligatures.
    2) All words must end in vowels - if a word has a consonant at the end
       without a vowel, it will be assumed that the "a" vowel is present.
       Thus, in hindi: instead of writing "hama", just "ham" is sufficient.
       This was done to increase readability. To get a viraam at the end
       of a word, use ".h" (without "a"), ex: "ham.h".
       This mechanism is activated for devanagari and bengali only.
    3) Added support for the the telugu font from the TeluguTeX package
    4) Removed support for the bit-mapped bengali font - BANGLA-V1.1
    5) Added support for the bit-mapped bengali font - SGAON
       This is originally a HP SoftFont, and can be converted to TeX PK 
       format, and to PostScript. Thus, it can be used with both TeX and
       PostScript.
    6) Devnac PostScript font: added the ligatures sha-ba, sha-va, sha-cha
    7) split up code in ichar.c to different files handling each indian
       language.
    8) added short marker symbols (!< and >!) to delimit indian lang texts
	(turned off by default, use command \useshortmarkers to enable).
    9) added support for digits (0..9) in the IFM file format.
    10) Default IFM file (dvnc.ifm) is not loaded  on startup, it will
       be loaded only if necessary --- so tamil only users will see
       faster startup times (as will telugu, bengali, etc users).
    11) Added the marathi la - (use mdvnc.ifm to get this).
       Changed the "gy" consonant to "GY" - to reduce conflict with
       the ga-ya consonant conjunct.
       Modified the dda-ya ligature.
       Improved the da-dha ligature.
       Many of the above changes/bug fixes to the devnac font have been made
       at the suggestion of Muduhakar Deshpande, I would like to thank him
       for his input.
    12) IFM file format has changed!

****** Version 2.11, 30 November 1991
    1) Minor changes, mainly break up package into two parts:
       itrns211 and itrnsps, where itrsps contains the printer
       ready postscript documentation. (ZIP or ZOO format archives.))
       This is in anticipation of downloading to IBM-PC archives, such
       as SIMTEL, etc.
    2) Fixed bug to allow \kern command in PostScript mode too..

****** Version 2.10, 14 October 1991
    0) Major features: IFM file added for Frans Velthuis's devanagari
       font (it is called Devnag);
       lots of new ligatures added to the Devnac PostScript font.
       Ligatures can now be inhibited by using the characters {}
       between two consonants in the input text.
       All ra-<consonants> can now be printed using the hindi form
       (before 2.1 these ligatures were not automatically handled,
        users had to use the special .r to get what they needed.
        From 2.1, the correct form will be used.)
       Marathi users can use alternate IFM files, which use the
       marathi form of ra-half.
    1) A few codenames (internal IFM file format data) changed to
       better reflect the actual character represented.
       This does mean that all pre 2.1 release itrans IFM files are now
       obsolete and incompatible with this and future releases.
       Also made a change in the way PCC .. dx dy works, for details
       see the file tech.tex.
    2) Finally have a better understanding of the archiac vowels (Ri, RI, Li
       and LI). Therefore the old naming scheme (RI RU LI RU) thrown out.
       RI in the old scheme is Ri in release 2.1
       RU in the old scheme is RI in release 2.1
       LI in the old scheme is Li in release 2.1
       LU in the old scheme is LI in release 2.1
       The Devnac font only had the Ri character, it was called
       RU earlier (pre 2.1).
       The Devnag font (Frans Velthuis's) uses all four of the above vowels.

--------------------------
****** Version 2.0, 26 August 1991
    0) Thanks to the following persons for many of the bug fixes and
	suggestions: Samir Agarwal, Madhu Deshpande, Gary Thrapp,
	Vick Khera, and many others.

    1) With the help of Sriram Tyagarajan, added support for Tamil.
       Uses the Washington Tamil font, it was created by the Humanities and
       Arts Computing Center of the University of Washington, USA.

    2) Because of 1, and other changes to support multi-languages,
       new keywords have been introduced:
       \indianifm=xxxxxxx   %% set the current IFM file
       There is the concept of a current IFM font file, which contains
       the font data used to construct indian language characters.
       All \marathi or \hindi or \tamil do is to mark the beginning of
       the transliterated text, they do not load in any IFM files, all
       loading must be done thru the \indianifm command.
       \indianfont=yyyyyyy %% command that will printed at \hindi
       See idoc.itex for more info.

       (Old documents with \marathi etc will still work since the default
	IFM file loaded in is devnac.ifm --- the hindi/marathi font
	metric file.)

    3) Changed file extension .itex to .itx, and changed name of file
       devnac.text to devnac.ps
       Lots of other changes, see idoc.itx the user's manual.

    4) added support for MSDOS machines (this version and onwards will
       be developed on the PC---a 80386 machine).
    5) Corrected the ifm definitions for big i, dha-ya ligature, etc
       in the devnac.ifm file.
    6) -I ./xxx.ifm would not work, it would still search for the
       file xxx.ifm in the ITRANSPATH.
       Fixed so that now it realizes that ./xxx.ifm is a full path
       name and  does not look for in the ITRANSPATH.
    7) Changed variable name "inline" to "line_in", some compilers
       did not like the word "inline" (shades of C++...) [font.c]

--------------------------
****** Version 1.01, 11 June 1991

    Most of these bugs came about to support non-ansi systems..

    1)  stdlib.h is missing on many systems, so removed its
	inclusion from itrans.h (instead, added the declaration
	for getenv() - the only thing needed from stdlib.h).

    2)  sprintf() works differently on non-ansi-C machines - instead of
	returning number of characters printed, it simply returns a
	pointer to the string!
	So, changed itotex.c and itops.c to specifically use strlen()
	instead of relying on sprintf().
	Thanks to Dinesh Dattatraya Gaitonde (CMU) for catching this.

    3) Bug in function font.c::S_ccadd() - return TRUE; missing, added.

    4) Specifically define the macro YYDEBUG so as to include the
       integer yydebug in iyacc.c

    5) YFLAGS was missing from the Makefile, added it.

--------------------------
