
	CHANGES IN THE PACKAGE
	----------------------

+ new feature
! bugfix
* other change

This list doesn't include some very unimportant changes and improvements.

---------------- version 1.8 (2 Sep 1998) ----------------

+ added `~\s ~' in *-tyldy filter

+ *-mnemonic1 filter; treatment of mnemonic1 changed from "filterm-only
  input filter" to "general konwert encoding", the same as mnemonic

* changed \(...\) into \{...\} in trs and UTF8-charset; this is because
  \(...\) together with \| will have other meaning in the future, and
  that meaning (multicharacter alternatives) is in all kinds of regexps
  marked with \(...\|...\) or (...|...); sorry for incompatibility

* removed \{...\} in trs; it was no longer used anywhere, can be
  emulated with \[...\], and we need \{...\} syntax for another purpose

! hmetodo now uses \? to properly handle multiple h's under filterm

+ setting of variables and rules conditionalized by variables in trs

! fixed treatment of reverse attribute in 512bold filter (thanks to Yann
  Dirson <ydirson@a2points.com>)

! fixed stupid bug in fixtrsutf8 which was giving bad results from rules
  with several sets of UTF-8 characters in them; rosyjski filter didn't
  contain such rule

! fixtrsutf8 produces slightly more optimized result, no longer
  enclosing single characters in \[...\]

+ Characters U+10000...U+7FFFFFFF supported by html{dec,hex} formats

+ 512bold now considers similarity of colors, throwing away not
  necessarily the least recently used color, but also sufficiently
  similar to the new color required; this slightly improves display when
  too many colors are wanted

! filters designed only for filterm documented in `man filterm' (were
  not described in man pages)

! ^[[...] Linux Console Private CSI Sequences implemented in 512bold

+ new 512bold filter which changes the palette dynamically: colors are
  not approximated, but when more than 8 are used at one time, some of
  them display incorrectly; old 512bold renamed into 512bold1, but it
  will not be developed, and at some time 512bold will be configurable
  to behave as 512bold1, which will then be dropped

! fixed treatment of ^[[0m and ^[8 in 512bold filter (thanks to Yann
  Dirson <ydirson@a2points.com>)

! fixed a typo in xmetodo-UTF8

+ rtf-8bit and 8bit-rtf (thanks to Ziemek Borowski
  <ziembor@FAQ-bot.Ziembor.waw.pl>)

! mkcharset-charset creates correct one-to-many tables when a
  character is represented in the source charset with several characters    
  looking the same as its approximation in the target charset, but the    
  single-character approximation is different (e.g. viqri-ascii)

+ filterm's option to execute a shell as a login shell (thanks to Yann
  Dirson <ydirson@a2points.com>)

! mnemonic and mnemonic1 formats updated, basing on glibc-2.0.7 sources

* characters 128...159 of ISO-8859-x mapped into U+0080...U+009F

* iso2cp1250utf8pl-iso2 removed; similar varpl-UTF8 added

+ vareo format

+ approximations of Vietnamese letters are based on viqri; with /vi
  option ascii output is the same as viqri

+ devel/hex-trs

+ viscii, viqri, and enhanced cp1258, which understands combined accents  
  (from Mark Leisher's <mleisher@crl.nmsu.edu> csets; thanks to Juliusz
  Chroboczek <jec@dcs.ed.ac.uk>)

! mkcharset-charset properly handles multiple-byte encoding

! empty trs argument is treated as a script rather than a file

+ language dependent translations of unavailable characters; mk*-charset
  rewritten again

* devel/fixmanconfig moved to aux/fixmanconfig to be placed in
  konwert-*.rpm rather than konwert-devel-*.rpm

* RPM package split into konwert and konwert-devel

---------------- version 1.7 (19 Jul 1998) ----------------

+ rmspacesateol filter

! added MIME charset name us-ascii

* rewritten mkUTF8-charset and mkcharset-charset to generate both tables
  for one-to-one and one-to-many mapping in one pass; mkUTF8-ascii is
  still run twice because I don't know how to specify one command making
  multiple targets in a Makefile; this change is not only a speed-up, but
  also prepares for future language dependent character substitutions

* When /etc/man.config uses -Tascii, during Konwert's installation it
  proposes fixing it to use -Tlatin1; use `make nofixmanconfig=1' when
  you don't want it, e.g. in packaging

+ Polish man pages

+ charset filters automatically use /1 option when used as output for
  filterm

+ alias macroman = mac

+ ISO-8859-15 and KOI8-RU (thanks to Juliusz Chroboczek <jec@dcs.ed.ac.uk>)

! man pages were not uninstalled by `make uninstall'

! fixed autodetection of two Esperanto encodings (there were typos in
  their names in any/eo)

! improved conversion of characters unavailable in the target charset -  
  now uses a suitable approximation of the replacement character U+FFFD
  found in UTF8-charset instead of using always `?'

! fixed unnecessary character substitutions listed in UTF8-charset
  filters for characters directly available (bug in mkUTF8-charset)

! fixed unnecessary one-to-one substitutions listed in charset-charset
  filters in one-to-many section (bug in mkcharset-charset)

+ trs guesses whether an argument without -f or -e is a filename or
  a script

+ `make rpm'

+ this CHANGES file, BUGS, and TODO

---------------- version 1.6 (4 Jul 1998) ----------------

+ RPM package (thanks to Ziemek Borowski <ziembor@FAQ-bot.Ziembor.waw.pl>)

! fixed UTF8-charset conversions when the set of characters that are    
  neither available in the target charset nor approximated with ASCII
  is empty

* changed UTF8-charset: Esperanto letters are now replaced with
  cx gx hx jx sx ux when unavailable directly

+ Esperanto xmetodo, hmetodo, antauxcxap, postcxap, apostrofoj,
  and malapostrofoj notations

+ filters can detect whether they are run by filterm

! htmlent format mangled some ASCII punctuation chars when used as
  output (thanks to Ziemek Borowski <ziembor@FAQ-bot.Ziembor.waw.pl>)

+ added iso3-ascii and iso10-ascii direct filters

+ ISO-8859-12,13,14 drafts

+ Added Zcaron and zcaron characters to recently changed CP-1252 (thanks
  to Adam Twardoch <adam.twardoch@euv-frankfurt-o.de>)

! fixed some language errors in English documentation (thanks to Gwidon
  S. Naskrent <naskrent@hoth.amu.edu.pl>)

+ Hebrew and Portugal encoding autodetection (thanks to Gwidon S.
  Naskrent <naskrent@hoth.amu.edu.pl>)

! fixed charset-charset conversions of characters not mentioned in  
  UTF8-charset, but available in both charsets (thanks to Gwidon S.
  Naskrent <naskrent@hoth.amu.edu.pl>)

+ English man pages (thanks to Gwidon S. Naskrent
  <naskrent@hoth.amu.edu.pl>) for konwert, trs and filterm, combining
  information from several doc/en/* files

+ option for showing the detected encoding is improved and documented

* charset tables moved from filters *-UTF8 into data files in
  aux/charsets to ease future understanding of options by these
  filters

* machine-independent files moved from $(prefix)/lib to $(prefix)/share
  (now they are all filters and their auxiliary files except 512bold)

+ charset aliases

---------------- version 1.5 (31 May 1998) ----------------

+ Debian package (thanks to Yann Dirson <ydirson@a2points.com>)

* scripts for calculating letter frequencies renamed from jakielitery
  and czestosci into whichletters and frequencies, and documented

* trs is no longer used during build (now it may be possible to
  cross-compile the package)

+ Greek, Esperanto, and Russian in &#Unicode_numbers; are also
  autodetected

* some files which take long time to build are already included in the
  distribution and not deleted by `make clean'; `make maintainer-clean'
  deletes them; this include UTF8-charset and charset-charset filters

! filterm fixed to compile and work properly under glibc

* #!/path substitution separated from the installation process, to ease
  future building of binary packages

---------------- version 1.4 (28 May 1998) ----------------

* #!/path substitution made with `perl -i' rather than `konwert '(trs)' -O',
  so it's a bit faster now

+ UTF-8 characters added to the iso2cp1250pl-iso2 filter; renamed into
  iso2cp1250utf8pl-iso2

! fixed html-UTF8 filter, which got broken in version 1.2

! Makefile now uses the perl specified in perl= option also during the
  build, not only for #!/path sunstitution

* Esperanto letter frequencies recalculated, basing on larger sample
  texts

+ scripts for calculating letter frequencies are included (but not used;
  they may be useful for other people wanting to contribute frequencies
  for autodetection of new languages)

* `CXXFLAGS=-Wall -O2 -fomit-frame-pointer -pipe' added to Makefile
  (thanks to Krzysztof G. Baranowski <kgb@manjak.knm.org.pl>)

+ pinectrlchars filter

+ expand and unexpand filters (using textutils' programs)

+ RFC-1541 mnemonics encoding (mnemonic)

+ updated devel/UTF8-charsets to find some one-character approximations

+ ability to specify character approximations for use only when some
  other characters are available (/1 option)

+ character-to-one-character-only option for charset conversions

+ kbdmacros and 512bold filters, and tyldy, lewyap, and boldpl formats
  for filterm

+ filterm program

* direct charset-charset filters for chosen pairs of charsets to  
  improve speed (built automatically)

---------------- version 1.3 (18 Apr 1998) ----------------

* a few changes in UTF8-charset and UTF8-tex tables

+ Greek, Esperanto, Spanish, French, Italian, and Swedish charset
  autodetection (thanks to Gwidon S. Naskrent <naskrent@hoth.amu.edu.pl>)

* English and Polish documentation moved into separate subdirs

! fixed some language errors in English documentation (thanks to Gwidon
  S. Naskrent <naskrent@hoth.amu.edu.pl>)

---------------- version 1.2 (14 Apr 1998) ----------------

* a few changes in UTF8-charset character approximation table

+ asciichar option for charset filters

* rosyjski conversion as an option for charset filters rather than as
  a standalone filter

+ general mechanism for adding filters to be applied to the intermediate
  UTF-8 form of charset conversion (aux/argcharset/*)

+ TeX method of representing extended characters; charset filters can
  use it for unavailable characters

* htmlchar format renamed into html

* internal intermediate U format for charsets renamed into UTF8

+ konwert finds a common intermediate format dynamically rather than
  searching for it according to a predefined list

---------------- version 1.1 (12 Apr 1998) ----------------

+ first public release

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

-- 
 __("<   Marcin Kowalczyk * qrczak@knm.org.pl http://qrczak.home.ml.org/
 \__/       GCS/M d- s+:-- a21 C+++>+++$ UL++>++++$ P+++ L++>++++$ E->++
  ^^                W++ N+++ o? K? w(---) O? M- V? PS-- PE++ Y? PGP->+ t
QRCZAK                  5? X- R tv-- b+>++ DI D- G+ e>++++ h! r--%>++ y-
