NEWS

liblouis NEWS -- history of user-visible changes.  	-*- org -*-

* Noteworthy changes in release 2.6.0 (2014-??-??)
** New features
- Added a callback system for logging purposes.

** Bug fixes
- Fixed bug to prevent removal of \xffff between largesign rules. This
  solves a LibLouisUTDML bug where \xffff is used as a segment delimiter.
- Fixed a bug in backtranslation, when a letsign was encountered, the
  letsign was being applied beyond the element it applied to.
- Fix memory leaks in the default table resolver introduced in the
  previous release.
- Fixes to the build system by Simon Aittamaa

** Braille table improvements
- Fix for Norwegian where letsign is affecting some extra characters
  thanks to Lars Bjørndal
- Much improved hyphenation for Norwegian thanks to Lars Bjørndal

* Noteworthy changes in release 2.5.4 (2014-3-3)
This release contains nine months worth of braille table improvements
for example for Danish, UEB, en-us, Nemeth, Bulgarian, Slovenian and
many more. Also there are fixes to the core for table path resolving
and back-translation.

** New features
   - Added function lou_registerTableResolver for plugging in a table
     resolver callback from your host language.

** Bug fixes
   - Fixed ENDSEGMENT indicator in computer Braille thanks to John
     Boyer.
   - Emphasized words should now be contracted consistently thanks to
     John Boyer.
   - Fixed several problems with back-translation. A slash within a
     number and strings such as 5-inch-diameter and 25-year-old-man
     should now back-translate correctly. Thanks to John Boyer.
   - Fixed a problem with syllable opcode. Thanks to John Boyer.
   - Fix warnings with gcc 4.8.2. Thanks to Peter Lundblad.
   - When a table is specified with an absolute or relative path, the
     "includes" in that table will now work as expected, meaning files
     in the same directory will be found.

** Braille table improvements
   - fr-bfu-comp8.utb: corrections, zero was wrongly displayed in some
     instances.
   - en-us-g2.ctb, en-ueb-g2.ctb: Fixes for that's, can't and s'
     thanks to Ken Perry.
   - en-us-g2.ctb: Fix for back-translation of things like
     http://address.com, words including after, capsigns. thanks to
     Ken Perry.
   - Further corrections and testcases to the Hungarian tables thanks
     to Hammer Attila
   - Fixed letter sign in en-us and en-ueb tables. Thanks to John
     Boyer and Ken Perry.
   - UEB Fixes thanks to Mesar Hameed, Ken Perry and Joseph Lee:
     - chardefs - correct title and fixed a long standing bug
       where dots 46 was inserted between letters (especially for web
       addresses).
     - Fix problem with at sign.
     - Removed section of accented letters, that were left behind from
       US table.
     - Added todo for accents, to define according to the formal docs.
     - Corrected mathematical forall symbol.
     - Modified emphasis dot combinations to follow UEB standard.
     - Corrected the display of period when used as a midword.
     - Ensure 'inin' is correctly back-translated in words such as
       asinine, feminine and others.
   - Bulgarian bg.ctb: updated to meet modern standards (added Latin
     letters, corrected punctuation/mathematical symbols, misc typos).
     Thanks to Rumiana Kamenska
   - en-gb-g1.utb: Fixes thanks to Paul Wood
   - Fix for the lich sign in the German tables. Thanks to Halim
     Sahin.
   - Nemeth improvements thanks to John Boyer.
     - Spaces in nemethdefs.cti were changed to unbreakable spaces.
       This was done so that Nemeth expressions would not be broken
       between lines.
     - The number sign is now inserted between the minus sign and the
       number at the beginning of an expression. Some problems with
       pass2 opcodes have also been corrected.
   - The Esperanto table has additional punctuation characters and a
     test harness. Thanks to Aaron Cannon.
   - Added missing symbols to the US English BRF display table. Thanks
     to Aaron Cannon.
   - Improvements to the Slovenian table and a new Slovenian eight dot
     computer table provided by Jožef Gregorc.
   - Fixes to the Norwegian tables thanks to Lars Bjørndal
     - Fixes for less than / greater than.
     - Add entries for URLs, domains and file names.
   - Added ne.utb, an alternative Nepali braille table. One of ne.utb
     or ne.ctb will be removed/merged in a future version. Keeping
     both for the time being so that users can test and give feedback
     on which is most correct. Thanks to Him Prasad Gautam, and Mesar
     Hameed
   - Much improved danish grade 1 and grade 2 6 and 8 dot braille,
     thanks to Bue Vester-Andersen and Mesar Hameed

** Backwards incompatible changes
   - Delete the table chardefs-ueb.cti as these rules are now provided
     by en-ueb-chardefs.uti

** Invisible changes
   - Refactoring in compileTranslationTable.c: separated more clearly
     the compilation from the table resolving, removed duplicate code,
     etc.

* Noteworthy changes in release 2.5.3 (2013-6-21)
** New features
*** New Braille tables
    - Korean grade 1 and grade 2 thanks to Joseph Lee
    - U.K. English 8 dot computer braille table thanks to David
      Reynolds
    - New Russian literary and computer braille tables thanks to Igor
      B. Poretsky. These replace the older Russian tables which are
      left for backwards compatibility.
    - New hyphenation dictionary for Russian thanks to Igor B.
      Poretsky.
    - Updated hyphenation tables for the Norwegian language (nynorsk
      and bokmål) thanks to Lars Bjørndal.
    - New hyphenation dictionary for Esperanto thanks to Aaron Cannon.
    - New Esperanto grade 1 table, using the x system for accented
      letters, thanks to Aaron Cannon.

*** runHarness.py: 
    Accept filename globs on the commandline to run specific harness files.
    In tests/harness, one can do make <filename> or make runall
    Removed from make check because these checks are checking the validity
    of our tables, rather than validity of the code.

** Braille Table Improvements
   - da-dk-g2.ctb, mostly rewritten to use nocross and hyphenation
     table.
   - Most tables: removed the default collapse whitespace statements,
     if you need to compress whitespace, consider adding compress.ctb
     to the list of tables when processing.
   - Corrections to Unified English Braille Code (Grade 1 and 2),
     thanks to Joseph Lee
   - Corrections to apostrophes in the Computer Spanish 8 dots Braille
     table (Es-Es-G0.utb). Thanks to Juan C. Buno.
   - Corrections for double angle quotation marks and emphasis marks
     in the Norwegian Grade 0 Braille Table. Thanks to Knut Arne
     Bjørndal.
   - Fixes for a minor problem regarding the noletsign in Norwegian
     contracted braille. Thanks to Lars Bjørndal.
   - Corrections to the Italian table thanks to Igor B. Poretsky.
   - Corrections to the Hungarian grade 1 table thanks to Hammer
     Attila
   - Corrections to English, U.S. Grade 2 (ABAE) table. Thanks to Ken
     Perry for reporting the bug and John J. Boyer for fixing it.
   - Further reorganization of the tables to remove duplication. Move
     litdigit opcode common parts and include where needed.
   - Removed obsolete en-us-g1.utb, which has been replaced by
     en-us-g1.ctb.
   - Added dictionary harness tests for: en-ueb-g2.ctb, en-us-g2.ctb.
   - Corrections to Nemeth character definitions thanks to Neil
     Soiffer.
   - Corrections to the Esperanto table thanks to Aaron Cannon.

** Bug Fixes
   - Cursor position calculation is now based on the same code that
     calculates inpos and outpos. This probably solves a number of
     bugs.
   - Fix nocross opcode processing.
   - Fix several buffer over/under runs in
     lou_translateString.c:hyphenate.
   - Fix the '=' problem, i.e. fix inputPositions calculation for the
     case where the equals sign is used as the dots operand. Thanks to
     Bert Frees
   - Fix a bug when resizing a table. Previously not all references to
     this table were updated.

** Backwards incompatible changes
   - The feature that allowed a mapping between language code and
     Braille table was removed as it contained a out-of-bounds access
     bug, was never documented and probably never used. Thanks to
     Peter Nilsson Lundblad and Jeremy Roman for analyzing this
     problem and providing a patch.

* Noteworthy changes in release 2.5.2 (2012-12-18)

While initially planned as mainly a bug fix release this release
contains some notable new features: There is a new tool to trace which
rules have been used to perform a translation. Also along with other
new tables the long awaited table for UEB is finally here.

** New features
*** New tool to trace rule application
    There is a new tool (lou_trace) which helps to trace which rules
    have been used to perform a Braille translation. This is helpful
    for writing Braille tables. See the documentation for more
    information.
*** New Braille tables
    - Inuktitut grade 1, thanks to Greg Kearney.
    - UEB grade 1 and 2, thanks to Joseph Lee. These tables replace
      the old UEB tables (UEBC-g1.utb and UEBC-g2.ctb).
    - Korean table thanks to Joseph Lee

** Braille Table Improvements
   - da-dk-g2.ctb, fixes for transposed â, å, æ, ä, ø  and ö,
     corrected/improved harness tests.
   - Corrections for en-GB-g2.ctb thanks to Paul Wood
   - Corrections to the Hungarian grade 1 table thanks to Hammer
     Attila

** Bug Fixes
   - Update gnulib
   - Fix a bug in the correct opcode which causes sometimes random
     results when translating. Thanks to Bert Frees.
   - Fixes for compiler warnings.
   - Fix some Valgrind warnings about invalid reads
 
* Noteworthy changes in release 2.5.1 (2012-9-24)

** Braille Table Improvements
- Fix encoding problem in italian table and added more character
  definitions. Thanks to Simone Dal Maso <simone.dalmaso@juvox.it>.
- Rename it-it-g1.utb to it-it-comp6.utb and it-it-g1.utb2 to
  it-it-comp8.utb.

** Bug fixes
- Fix outputPos and inlen where an input character generates multiple
  output characters.

* Noteworthy changes in release 2.5.0 (2012-9-10)

This release contains a tremendous amount of work many developers.
Many long standing bugs have been fixed. The tables can finally be in
UTF-8. A grand table cleanup removed duplication from the tables.
There are now two extensive test frameworks for table writers. A
number of new tables have been contributed on top of the usual
assortment of table improvements. Thanks to all of this liblouis has
already seen quite a bit of uptake in a number of places, notably the
new DAISY pipeline will ship with this release of liblouis.

NOTE: If you have private tables you might want to migrate them to
utf-8. To do this just use iconv as follows:

  $ iconv -f latin-1 -t utf-8 <input >output

** New features
*** New Braille tables
    - Estonian grade 0, thanks to Jürgen Dengo.
    - Portuguese 8 dot Computer braille, Thanks to Rui Fontes
*** UTF-8 support in tables
    Braille tables can now contain UTF-8 in the opcode arguments.
*** Improvements to the python bindings
    All constants defined in liblouis.h are now exposed in the bindings.
*** Add a doctest infrastructure
    These tests are based on the Python doctest framework and are only
    run if there is a Python interpreter on the system
*** Add a test harness
    This test infrastructure allows the user to do table tests in a
    simple and concise syntax. These tests are based on the Python
    nose testing framework and are only run if either Python 2.x or
    3.x with the related nose python module is installed on the
    system. See the documentation for more information. Thanks to
    Mesar Hameed.
*** Add a test harness generator
    A harness generator that uses simple text files with a little
    formatting to help to generate the json harness files. The purpose
    of this tool is to make it much easier and faster to add checks
    for a given table. You are expected to read the generated harness
    file and make necessary changes, the tool only helps you to get
    the tests into the harness format, not check their validity.
*** Support for Python 3 in the Python bindings
    The Python bindings now work for both Python 2 and Python 3.
    Thanks to Michael Whapples.

** Improved C-based test framework
   - Improved the test framework to be able to test translations
     involving Unicode.
   - Added numerous tests, e.g. for lowercase and Unicode, for the
     input position, for repeated, etc.

** Improved the documentation
   - Document the test harness (json format, fields, flags).
   - Document the use of Valgrind to find memory leaks
   - Improve the documentation on the display opcode

** Bug fixes
   - lou_allround and lou_translate now properly handle Unicode
     characters
   - Fix some issues reported by Valgrind
   - Fix inputPos for situation where context and multipass opcodes
     are involved
   - Fixed a number of bugs with the letter, uppercase and lowercase
     opcodes when dealing with Unicode
   - Fixed a couple of bugs with hyphenation (documentation, Python
     bindings and a number of buffer overruns in the C library).
     Thanks Milan Zamazal <pdm@brailcom.org> for reporting this.
   - Fix a bug in the $a. matcher in the multipass rules where only 32
     chars were matched. It now matches 0xffff chars.
   - Fix a bug reported by James Teh related to pass1Only

** Braille Table Improvements
   - all table files have consistent encoding, UTF-8.
   - The grand table cleanup: Reorganize the tables to remove
     duplication. Move common parts such as Latin letter, eight and
     six dot digit definitions to separate files which are then
     included. This should ease table maintenance. Thanks to Mesar Hameed. 
   - Fixes to de-de-comp8.ctb thanks to Aliminator83@gmail.com
   - hu1.ctb renamed to hu-hu-g1.ctb
   - hu.ctb renamed to hu-hu-comp8.ctb
   - eo.ctb renamed to eo-g1.ctb
   - Fixes to eo-g1.ctb thanks to Aaron Cannon <cannona@fireantproductions.com>
   - hu-hu-g1.ctb: improvements and extensive test harness, with
     working back-translation, Thanks to Hammer Attila
   - Fixes to fr-bfu-comp6.utb and fr-bfu-comp8.utb thanks to Michel
     Such <michel.such@free.fr>
   - Reworked and extended Ethiopic braille table ethio-g1.ctb,
     superseeds gez*, thanks to Dr. Tamru E. Belay
     <g.braille@sympatico.ca>
   - Fixes to no-no-g3.ctb thanks to Lars Bjørndal <lars@lamasti.net>

* Noteworthy changes in release 2.4.1 (2012-2-22)

** New features
   - Czech hyphenation table thanks to Jan Hegr
   - Spanish grade 1 table provided by José Enrique Fernández del
     Campo and Juan Carlos Buño Suárez
   - New Tamil table thanks to Mesar Hameed

** Braille Table Improvements
   - Improvements to the Portuguese grade1 braille tables
   - Updates and additions to Icelandic 8-dot braille table.
   - Improvements to the uncontracted Spanish computer braille table.
   - Improvements to the Norwegian braille table thanks to David Hole.

* Noteworthy changes in release 2.4.0 (2012-01-31)

** New features
   - New Generic Farsi Grade 1 table: A new table for Generic Farsi
     Grade 1 braille has been provided by Mesar Hameed.
   - Emacs mode for editing Braille tables thanks to Christian Egli

** Braille Table Improvements
   - Improvements to the French comp6 and comp8 braille tables
   - Improvements to the Romanian braille table
   - Improvements to the Generic Arabic Grade 1 table
   - Improvements to the Czech tables thanks to Jan Halousek and to
     Jan Hegr

* Noteworthy changes in release 2.3.0 (2011-05-09)

This release contains support for many more languages than before
(Swedish, Kurdish, Ethiopic, Serbian, many Indian languages). The
search path for tables is now a list of paths. Finally there is the
usual assortment of bug fixes.

** New features
*** Multiple table search path
    The environment variable LOUIS_TABLEPATH can now contain a list of
    paths (separated by commas) where liblouis should look for tables.
    This allows the user to keep local tables.
*** New --quiet option for lou_checktable
    lou_checktable writes to stderr even in the case of success. This
    can now be suppressed with the new option --quiet.
*** New Swedish table
    A new table for Swedish braille has been provided by Samuel
    Thibault.
*** New table for Sorani (Kurdish)
    A new table for Sorani (Kurdish) Braille has been donated by Peter
    Engström from Index Braille
*** New table for Ethiopic
    A new table for Ethiopic Braille has been donated by Tamru E.
    Belay PH.D from Adaptive Technology Center for the Blind (ATCB)
*** New table for Serbian
    A new table for Serbian Braille has been donated by Peter Engström
    from Index Braille

** Improved the documentation
   The deprecated opcodes have been moved to a separate section

** Bug fixes
   - Fixed a long standing bug with an infinite loop in the table
     compiler

** Braille Table Improvements
   - Improvements to the Chinese braille table
   - Improvements to the Flemish Braille Math Code tables
   - Improvements to the Dutch	Braille tables
   - Improvements to the Spanish Braille tables.
   - Fixes for the uncontracted French 6 and 8 dot tables
   - Improved support for Italian 8 dot
   - Improvements to the Generic Arabic Grade 1 table

* Noteworthy changes in release 2.2.0 (????-??-??)

** New features
*** New tables
    - Support for many indian languages
    - Support for Icelandic 6- and 8-dot
    - Support for Catalan
    - Support for Dutch Braille (for Belgium and the Netherlands)
    - Support for Flemish Braille Math Code (a.k.a. Woluwe code)


*** New functions to make libraries relocatable
    Two new functions, to set the search path for tables and files.
    They make the library relocatable. See the in the documentation
    for lou_setDataPath and lou_getDataPath.

** Bug fixes
    - Improved support for Spanish
    - Improved Norwegian tables

* Noteworthy changes in release 2.1.1 (2010-8-23)

** Bug fixes
   - Fixed problems with the Danish grade 2 table
   - Fixed problems with the Marburg maths table for mathematics and
     the UK maths table for mathematics

* Noteworthy changes in release 2.1.0 (2010-8-19)

** New features
*** New tables
    - Added tables for Portuguese grade 1 and 2
    - Added unicode.dis for Unicode braille
*** Modified tables
    - Updated Danish tables
*** Implemented language to table mapping
*** New format of error messages
    The error messages are now reported in a format similar to the one
    used in gcc. 
*** New opcode
    - added undefined opcode
*** Python bindings
    - Allow the user to configure the maximum output length by
      specifying a number by which the input length is multiplied
      using the outlenMultiplier module variable. The default will
      handle the case where every input character is undefined in the
      translation table. Previously, this was hard-coded to 2, which
      was insufficient in some cases.
    - Add compbrlLeftCursor mode constant.
    - Add compileString function which wraps lou_compileString.
    - Corrections/clarifications to docstrings.
    - Add python binding for the lou_hyphenate function.
    - Added python wrapper for lou_backTranslateString and lou_backTranslate.

*** liblouisxslt as an example
    Add liblouisxslt as an example to python/examples. This is
    basically an extension of libxslt that lets you invoke liblouis
    from an xslt stylesheet to do Braille translation on text nodes
    for example. 
*** compbrlLeftCursor
    Added a patch provided by Volker Bijewitz to implement
    compbrlLeftCursor.

** Bug fixes
*** output cursorPos
    Fix the output cursorPos when the compbrlAtCursor mode is enabled
    and the characters around the cursor translate to multiple braille
    cells, such as in the Chinese braille tables.
*** outpos when doing back translation
    Include a patch by Timothy Lee to fix outpos when doing back translation
      (issue 11)
*** inputPos/outputPos for undefined characters
    Fix the input/output position arrays for characters in the input
      which are undefined in the translation table.
*** table fixes
    - Fixed a bug with back translation of '*n'. (issue 13)
    - Fixes to the en-us-g2.ctb table
*** Python bindings
    - Remove unnecessary imports, allowing the bindings to run in Python
      2.7. (issue 12)
    - lou_translate* writes output information in typeform, so
      allocate enough bytes for it. Fixes possible buffer overruns and
      resultant crashes.
*** Miscellaneous
    - Fixes to the man page generation to fix issues that were
      reported by the Debian packaging builder
    - Do not invoke help2man when cross-compiling 
    - Documentation updates (issue 10)
    - Removing noletsign defaults
    - Many small fixes

* Noteworthy changes in release 2.0.0 (2010-7-6)

** New features
*** New functions
    - Adding lou_charSize function

* Noteworthy changes in release 1.9.0 (2010-6-29)

** New features
*** New functions
    - lou_dotsToChar and lou_charToDots function
    - Added lou_compileString for adding entries to tables at
      run-time.

* Noteworthy changes in release 1.8.0 (2009-11-23)

This release contains a number of improvements notably the integration
of gnulib, the automatic generation of man pages and the addition of
tables for German grade 2.

** New features
*** New tables
    - Tables German Grade 2
    - Swiss German
    - Swedish (1989 standard)
    - Swedish (1996 standard)

*** Modified tables
    - Updated Norwegian tables
    - Updated Chinese braille table

*** man pages
    All tools accept the --version and --help options and are
    documented in man pages

*** Corpus based test cases for tables
    You can now have corpus based tests for tables. See the README in
    tests/table_test_corpuses.

** Bug fixes
   - config.h is no longer exported
   - Many small fixes

* Noteworthy changes in release 1.7.0 (2009-08-21)

The main new feature of this release is the support for UK and Marburg
math. Other changes include a new tool to check hyphenation and the
usual improvement and addition of tables. Also The test suite has been
enhanced and finally passes.

** New features
*** New tables
    - Tables for UK and Marburg math
    - Hong Kong Cantonese
    - Hebrew
    - Hungarian
    - Slovene
    - Tibetan
    - Irish
    - Maltese

*** Modified tables
    - Updated Norwegian tables
    - Bug fixes in Russian tables
    - Updated French tables

*** lou_checkhyphens tool
    New tool to check hyphenation
*** rpm spec file
*** Test cases for tables
    The tables can now be tested with `make check'
*** New opcodes
    - noback and nofor opcode prefixes
    - grouping opcode
    - multipass subopcodes

** Bug fixes
   - Fix for library name and Python bindings
   - Documentation fixes
   - Many small fixes

* Noteworthy changes in release 1.6.2 (2009-05-01)

This release contains a new opcode for Malaysian Braille. See the
documentation for a description of the new opcode.

** New features

*** repword opcode 
The repword opcode is needed for Malaysian Braille

* Noteworthy changes in release 1.6.1 (2009-04-21)

This is mostly a bug fix release. It contains many bug fixes that were
discovered in the course of developing UK Math tables.

** Bug fixes

*** bug fixes for correct, context and multipass opcodes
*** bug fixes for largesign opcode
*** fixed bug with French back-translation
*** fixed the installation path for docs
*** documentation improvement

* Noteworthy changes in release 1.6 (2009-03-04)

This release features support for Danish and Russian and updated
tables for French and Norwegian. The search path for tables can now be
specified using an environment variable. Finally there is the usual
assortment of bug fixes.

** New features

*** exactdots opcode
The exactdots opcode is intended for use in liblouisxml
semantic-action files to specify exact dot patterns, as in
mathematical codes.
*** LOUIS_TABLEPATH env variable
You can now specify where liblouis is to look for tables with the
LOUIS_TABLEPATH environment variable.
*** New Tables for Danish and Russian
There is now support for Danish and Russian.

** Bug fixes

*** Updated French and Norwegian tables
*** Use stdcall calling convention if building for Windows

** Changes in behavior

None

* Noteworthy changes in release 1.5 (2009-01-21)

This is a big release for liblouis. It's the first time that it is
done from the Google code page. A number of people have contributed,
namely John Boyer (table debugger, bug fixes), Eitan Isaacson (Python
bindings), James Teh (Python bindings, bug fixes), Christian Egli
(documentation) and Michel Such (table for French grade 2).


** New features

*** Python bindings
The liblouis library can now be used from Python. For more info
consult the README file in the python directory.

*** Table debugger
liblouis now comes with a debugger that can help to find problems with
translation tables.

*** French table for grade 2
There is now a translation table for French grade 2.

*** pass1Only mode bit
The new pass1Only mode bit will help developers of screen readers as
the cursor will stay where it is expected to.

** Bug fixes

Fix the inpos array values for the case where a rule has an output
length which is larger than its input length.

fixed multi-word phrases

fixed bug in character display

fixed bug in findOpcodeName

lou_version now returns the correct liblouis version

** Changes in behavior

None

#+OPTIONS: toc:nil num:nil