The Linux Cyrillic HOWTO
  Alexander L. Belikoff, (abel@bfr.co.il), Berger Financial
  Research Ltd.
  v4.0, 23 January 1998

  This document describes how to set up your Linux box to typeset, view
  and print the documents in the Russian language.
  ______________________________________________________________________

  Table of Contents
























































  1. Administrativia

     1.1 Introduction
     1.2 Availability and feedback
     1.3 Acknowledgments and copyrights

  2. Theoretical background

     2.1 Characters and codesets

  3. Preparing your environment

     3.1 Text mode setup
        3.1.1 Linux Console
        3.1.2 FreeBSD Console
     3.2 The X Window System
        3.2.1 The X fonts.
        3.2.2 The input translation
     3.3 First steps - Cyrillic in shells
     3.4 bash
     3.5 csh/tcsh
     3.6 ksh
     3.7 less
     3.8 mc (The Midnight Commander)
     3.9 rlogin
     3.10 zsh

  4. Editing text

     4.1 Emacs and XEmacs
     4.2 Using vi
     4.3 Editing text with joe
     4.4 Spell-checking Russian

  5. Using Cyrillic with mail and news

     5.1 Setting up Mail User Agents
        5.1.1 Emacs-based mail readers
        5.1.2 pine
     5.2 Configuring your MTA
        5.2.1 sendmail
        5.2.2 Other MTAs

  6. Browsing the Cyrillic Web

     6.1 lynx
     6.2 Netscape navigator
        6.2.1 Basic setup
        6.2.2 Cyrillic text in frames and input areas
        6.2.3 Advanced setup

  7. Cyrillic wordprocessing

     7.1 TeX-based environments
        7.1.1 Using the Washington Cyrillic
        7.1.2 KOI-8 package for teTeX
        7.1.3 Using the cmcyralt package for LaTeX
        7.1.4 Using the CyrTUG package
     7.2 The StarOffice suite

  8. Printing and PostScript

     8.1 Text to PostScript conversion
        8.1.1 An a2ps converter
        8.1.2 The GNU enscript
     8.2 Text to TeX conversion
  9. Cyrillic in PostScript

     9.1 Adding Cyrillic fonts to Ghostscript

  10. Print setup

     10.1 Pre-loading Cyrillic fonts into a non-PostScript printer
     10.2 Printing with different fonts

  11. Localization and Internationalization

     11.1 Locale
        11.1.1 How to use locale
        11.1.2 Locale-aware programming
     11.2 Internationalization

  12. Staying compatible

     12.1 MIME-based data compatibility
     12.2 Explicit character set conversion
     12.3 Cyrillic in the DOS emulator

  13. Bibliography

  14. Summary of the various useful resources



  ______________________________________________________________________

  1.  Administrativia



  1.1.  Introduction


  This document covers the things you need to successfully work with
  information containing cyrillic text (mostly Russian) under Linux.
  Although this document assumes your using Linux as an operating
  system, most of information presented is equally applicable to many
  other Unix flavors. I shall try to keep the distinction as visible as
  possible.

  There are a number of popular Linux distributions. As an example
  system I describe the RedHat 4.1 Linux (Vanderbildt) - the one I am
  personally using. Nevertheless, I shall try to highlight the
  differences, if they exist, in other popular distributions, such as
  Debian GNU/Linux and Slackware Linux.

  Since such setup directly modifies and extends the Operating System,
  you should understand, what you are doing. Even though I tried to keep
  things as easy as possible, having some experience with a given piece
  of software is an advantage. I am not going to describe what the X
  Window System is or how to typeset the documents with TeX and LaTeX,
  or how to install printer in Linux. Those issues are covered in other
  documents.

  For the same reason, in most cases I describe a system-wide setup, by
  default requiring root privileges. Still, if there is a possibility
  for user-level setup, I'll try to mention it.

  NOTE: The X Window System, TeX and other Linux components are complex
  systems with a sofisticated configuration. If you do something wrong,
  you can not only fail with Russian setup, but to break the component
  as well, if not the entire system. This is not to scare you off, but
  merely to make you understand the seriousness of the process and be
  careful. Preliminary backup of the config files is highly recommended.
  Having a guru around is also advantageous.




  1.2.  Availability and feedback


  This document is available at sunsite.unc.edu or tsx-11.mit.edu as a
  part of the Linux Document Project. Also, it may be available at
  various FTP sites containing Linux. Moreover, it may be included as a
  part of Linux distribution.

  If you have any suggestions or corrections regarding this document,
  please, don't hesitate to contact me as abel@bfr.co.il. Any new and
  useful information about Cyrillic support in various Unices is highly
  appreciated. Remember, it will help the others.



  1.3.  Acknowledgments and copyrights


  Many people helped me (and not only me) with valuable information and
  suggestions. Even more people contributed software to the public
  community. I am sorry if I forgot to mention somebody.

  So, here they go:


  o  Bas V. de Bakker

  o  David Daves

  o  Serge Vakulenko

  o  Sergei O. Naoumov

  o  Winfried Truemper

  o  Ilya K. Orehov

  o  Michael Van Canneyt

  o  Alex Bogdanov

  o  ...and the countless helpful people from the relcom.fido.ru.unix
     and relcom.fido.ru.linux Usenet newsgroups.

  This document is Copyright (C) 1995,1997 by Alexander L. Belikoff. It
  may be used and distributed under the usual Linux HOWTO terms
  described below.

  The following is a Linux HOWTO copyright notice:


       Unless otherwise stated, Linux HOWTO documents are copy-
       righted by their respective authors. Linux HOWTO documents
       may be reproduced and distributed in whole or in part, in
       any medium physical or electronic, as long as this copyright
       notice is retained on all copies. Commercial redistribution
       is allowed and encouraged; however, the author would like to
       be notified of any such distributions.

       All translations, derivative works, or aggregate works
       incorporating any Linux HOWTO documents must be covered
       under this copyright notice.  That is, you may not produce a
       derivative work from a HOWTO and impose additional restric-
       tions on its distribution. Exceptions to these rules may be
       granted under certain conditions; please contact the Linux
       HOWTO coordinator at the address given below.



       In short, we wish to promote dissemination of this informa-
       tion through as many channels as possible. However, we do
       wish to retain copyright on the HOWTO documents, and would
       like to be notified of any plans to redistribute the HOWTOs.


  If you have questions, please contact Tim Bynum, the Linux HOWTO
  coordinator, at linux-howto@sunsite.unc.edu. You may finger this
  address for phone number and additional contact information.

  Unix is a technology trademark of the X/Open Ltd.; MS-DOS, Windows,
  Windows 95, and Windows NT are trademarks of the Microsoft Corp.; The
  X Window System is a trademark of The X Consortium Inc. Other
  trademarks belong to the appropriate holders.



  2.  Theoretical background



  2.1.  Characters and codesets


  In order to understand and print characters of various languages, the
  system and software should be able to distinguish them from other
  characters. That is, each unique character must have a unique
  representation inside the operating system, or the particular software
  package. Such collection of all unique characters, that the system is
  able to represent at once, is called a codeset.

  At the time of the most operating system's creation, nobody cared
  about software being multilingual. Therefore, the most popular codeset
  was (and actually is) an ASCII (American Standard Code for Information
  Interchange).

  The standard ASCII (aka 7-bit ASCII) comprises 128 unique codes. Some
  of them ASCII defines as real printable characters, and some are so-
  called control characters, which had special meanings in the old
  communication protocols. Each element of the set is identified by an
  integer character code (0-127). The subset of printable characters
  represents those found on the typewriter's keyboard with some minor
  additions. Each character occupies 7 least significant bits of a byte,
  whereas the most significant one was used for control purposes (say,
  transmission control in old communication packages).

  The 7-bit ASCII concept was extended by 8-bit ASCII (aka extended
  ASCII). In this codeset, the characters' codes' range is 0-255. The
  lower half (0-127) is pure ASCII, whereas the upper one contains 127
  more characters. Since this codeset is backward compatible with the
  ASCII (character still occupies 8 bit, the codes correspond the old
  ASCII), this codeset gained wide popularity.

  The 8-bit ASCII doesn't define the contents of the upper half of the
  codeset. Therefore the ISO organization took the responsibility of
  defining a family of standards known as ISO 8859-X family. It is a
  collection of 8-bit codesets, where the lower half of each codeset
  (characters with codes 0-127) matches the ASCII and the upper parts
  define characters for various languages. For example, the following
  codesets are defined:


  o  8859-1 - Europe, Latin America (also known as Latin 1)

  o  8859-2 - Eastern Europe

  o  8859-5 - Cyrillic

  o  8859-8 - Hebrew

  In Latin 1, the upper half of the table defines various characters
  which are not part of the English alphabet, but are present in various
  european languages (german umlauts, french accentes etc).

  Another popular extended ASCII implementation is so-called IBM
  codepage (named after some computer company, that developed this
  codeset for it's infamous personal computers). This one contains
  pseudo-graphic characters in the upper half.

  Software, that doesn't make any assumptions about the 8-th bit of the
  ASCII data is called 8-bit clean. Some older programs, designed with
  7-bit ASCII in mind are not 8-bit clean and may work incorrectly with
  your extended ASCII data. Most of packages, however, are able to deal
  with the extended ASCII by default, or require some very basic setup.
  NOTE: before posting the question "I did all setup right, but I cannot
  enter/view Cyrillic characters!", please consult the section ``'' for
  the notes on the program, you are using.

  For information about making your software 8-bit clean, see section
  ``''.

  Since on most systems character occupies 8 bits, there is no way to
  extend ASCII more and more. The way to implement new symbols in ASCII-
  based codesets is creation of other extended ASCII implementations.
  This is the way, the Cyrillic ASCII set is implemented.

  We already mentioned ISO 8859-5 standard as the one defining the
  Cyrillic codeset. But as it often happens to the standards, this one
  was developed without taking into account the real practices in the
  former USSR. Therefore, one thing that standard really achieved was
  another degree of confusion. I wouldn't say that ISO 8859-5 is widely
  used anywhere.

  Other standards for Cyrillic include the so-called Alt codeset and
  Microsoft CP1251 codepage. The former one was developed by (who?) for
  MS-DOS quite a while ago. Back then, there was not very buzz yet about
  internetworking, so the intention was to make it as compatible as
  possible with the IBM standard. Therefore the Alt codeset is
  effectively the same IBM codepage, where all specific European
  characters in the upper half were replaced with the Cyrillic ones,
  leaving the pseudographic ones. Therefore, it didn't screw the text
  windowing facilities and provided Cyrillic characters as well.  The
  Alt standard is still alive and extremely popular in MS-DOS.

  Microsoft CP1251 codepage is just an attempt of Microsoft to come up
  with the new standard for Cyrillic codeset in Windows. As far as I
  know, it is not compatible with anything else (not very surprizing,
  huh?)

  And finally there is KOI8-R. This one is also quite old, but it was
  designed wisely and nowadays the design points of it look really
  useful.
  Again, it is compatible with ASCII, and the Cyrillic characters are
  located in the upper half. But the main design point of KOI8-R is that
  the Cyrillic characters' positions must correspond to the English
  characters with the same phonetics. Namely, if we set the eighth bit
  of the English character 'a', we'll get the Cyrillic 'a'.  This means
  that, given the Cyrillic text written in KOI8-R, we can strip the
  eighth bit of each character and we still get a readable text,
  although written with English characters! This is very important now,
  since there are many mailers on the Internet, that just strip the
  eighth bit silently, being sure that every single soul on the face of
  the Earth speaks English.

  Not surprisingly, KOI8-R quickly became a de-facto standard for
  Cyrillic on the Internet. Andrew A. Chernov did a tremendous amount of
  work to make a standard in this area. He is an author of RFC 1489
  ("Registration of a Cyrillic Character Set").

  These two standards differ only in positions of the cyrillic
  characters in the table (that is in cyrillic character codes).

  The principal difference is that the Alt codeset is used by MS-DOS
  users only, whereas KOI8-R is used in Unix, as well as in MS-DOS
  (though in the latter KOI8-R is much less popular). Since we are doing
  the right thing (namely working in the Unix operating system), we
  shall focuse mostly on KOI8-R.

  As for the ISO standard, it is more popular in Europe and the US as a
  standard for Cyrillic. The leader in Russia is definitely KOI8-R.

  There are other standards, which are different from ASCII and much
  more flexible. Unicode is most known. However, they are not
  implemented as good as the basic ones in Unix in general and  Linux in
  particular. Therefore, I am not describing them here.



  3.  Preparing your environment


  Before we start customizing various parts of the system functionality,
  we have to set up a couple basic things. Most of tools described below
  assume that there are Cyrillic fonts available and a user is able to
  input Cyrillic characters. To make it true we have to configure the
  environment to provide both fonts and input facility for Cyrillic.

  There are effectively two interface models supported by Linux. One is
  the text mode, and the other one is the graphic mode, provided by the
  X Window System. Both require different setup, which will be described
  below.



  3.1.  Text mode setup


  Generally, the text mode setup is the easiest way to show and input
  Cyrillic characters. There is one significant complication, however:
  the text mode fonts and keyboard layout manipulations depend on
  terminal driver implementation. Therefore, there is no portable way to
  achieve the goal across different systems.

  Right now, I describe the way to deal with the Linux console driver.
  Thus, if you have another system, don't expect it to work for you.
  Instead, consult your terminal driver manual. Nevertheless, send me
  any information you find, so I'll be able to include it in further
  versions of this document.
  3.1.1.  Linux Console


  The Linux console driver is quite a flexible piece of software. It is
  capable of changing fonts as well as keyboard layouts. To achieve it,
  you'll need the kbd package. Both RedHat and Slackware install kbd as
  part of a system.

  The kbd package contains keyboard control utilities as well as a big
  collection of fonts and keyboard layouts.

  Cyrillic setup with kbd usually involves two things:


  1. Screen font setup. This is performed by the setfont program. The
     fonts files are located in /usr/lib/kbd/consolefonts.

     NOTE: Never run the setfont program under X because it will hang
     your system. This is because it works with low-level video card
     calls which X doesn't like.

  2. Load the appropriate keyboard layout with the loadkeys program.

  NOTE: In RedHat 3.0.3, /usr/bin/loadkeys has too restrictive access
  permissions, namely 700 (rwx------). There are no reasons for that,
  since everyone may compile his own copy and execute it (the
  appropriate system calls are not root-only). Thus, just ask your
  sysadmin to set more reasonable permissions for it (for example, 755).

  The following is an excerpt from my cyrload script, which sets up the
  Cyrillic mode for Linux console:


  if [ notset.$DISPLAY != notset. ]; then
      echo "`basename $0`:  cannot run under X"
      exit
  fi

  loadkeys /usr/lib/kbd/keytables/ru.map
  setfont /usr/lib/kbd/consolefonts/Cyr_a8x16
  mapscrn /usr/lib/kbd/consoletrans/koi2alt
  echo -ne "\033(K"              # the magic sequence
  echo "Use the right Ctrl key to switch the mode..."



  Let me explain it a bit. You load the appropriate keyboard mapping.
  Then you load a font corresponding to the Alt codeset. Then, in order
  to be able to display text in KOI8-R correctly, you load a screen
  translation table. What it does is a translation of some characters
  from the upper half of the codeset to the Alt encoding. The word
  'some' is crucial here - not all characters get translated, therefore
  some of them, like IBM pseudographic characters get unmodified to the
  screen and display correctly, since they are compatible with the Alt
  codeset, as opposed to KOI8-R. To ensure this, run mc and pretend you
  are back to MS-DOS 3.3...

  Finally, the magic sequence is important but I have no idea what on
  the Earth it does. I stole/borrowed/learned it from German HOWTO back
  in 1994, when it was like the only national language oriented HOWTO.
  If you have any idea about this magic sequence, please tell me.

  Finally, for those purists, who don't wont to give the Alt codeset a
  chance, I'm attaching yet another version of the script above, using
  native KOI8-R fonts.

  if [ notset.$DISPLAY != notset. ]; then
      echo "`basename $0`:  cannot run under X"
      exit
  fi

  loadkeys /usr/lib/kbd/keytables/ru.map
  setfont /usr/lib/kbd/consolefonts/koi-8x16
  echo "Use the right Ctrl key to switch the mode..."



  However, don't expect nice borders in your text mode-based windowing
  applications.

  Now you probably want to test it. Do the appropriate bash or tcsh
  setup, rerun it, then press the right Control key and make sure you
  are getting the cyrillic characters right. The 'q' key must produce
  russian "short i" character, 'w' generates "ts", etc.

  If you've screwed something up, the very best thing to do is to reset
  to the original (that is, US) settings. Execute the following
  commands:


  loadkeys /usr/lib/kbd/keytables/defkeymap.map
  setfont /usr/lib/kbd/consolefonts/default8x16



  NOTE: unfortunately enough, the console driver is not able to preserve
  it's state (at least easily enough), while running the X Window
  System. Therefore, after you leave the X (or switch from it to a
  console), you have to reload the console russian font.



  3.1.2.  FreeBSD Console


  I am not using FreeBSD so I couldn't test the following information.
  All data in this section should be treated as just pointers to begin
  with. The FreeBSD project homepage may have some information on the
  subject. Another good source is the relcom.fido.ru.unix newsgroup.
  Also, check the resources listed in section ``''.

  Anyway, this is what Ilya K. Orehov suggests to do in order to make
  FreeBSD console speak Russian:


  1. In /etc/sysconfig add:



     keymap=ru.koi8-r
     keyrate=fast
     # NOTE: '^[' below is a single control character
     keychange="61 ^[[K"
     cursor=destructive
     scrnmap=koi8-r2cp866
     font8x16=cp866b-8x16
     font8x14=cp866-8x14
     font8x8=cp866-8x8




  2. In /etc/csh.login:


     setenv ENABLE_STARTUP_LOCALE
     setenv LANG ru_SU.KOI8-R
     setenv LESSCHARSET latin1



  3. Make analogous changes in /etc/profile



  3.2.  The X Window System


  Like the console mode, the X environment also requires some setup.
  This involves setting up the input mode and the X fonts. Both are
  being discussed below.



  3.2.1.  The X fonts.


  First of all, you have to obtain the fonts having the Cyrillic glyphs
  at the appropriate positions.

  If you are using the most recent X (or XFree86) distribution, chances
  are, that you already have such fonts. In the late 1995, the X Window
  System incorporated a set of Cyrillic fonts, created by Cronyx. Ask
  your system administrator, or, if you are the one, check your system,
  namely:


  1. Run 'xlsfonts | grep koi8'. If there are fonts listed, your X
     server is already aware about the fonts.

  2. Otherwise, run


     find -name crox\*.pcf\*



  to find the location of the Cyrillic fonts in the system. You'll have
  to enable those fonts to the X server, as I explain below.

  If you haven't found such fonts installed, you'll have to do it
  yourself.

  There is some ambiguity with the fonts. XFree86 docs claim that the
  russian fonts collection included in the distribution is developed by
  Cronyx. Nevertheless, you may find another set of Cronyx Cyrillic
  fonts on the net (eg. on ftp.kiae.su), known as the xrus package
  (don't confuse it with the xrus program, which is used to setup a
  Cyrillic keyboard layout. Hopefully, tha letter one was renamed to
  xruskb recently). Xrus has fewer fonts than the collection in Xfree86
  (38 vs 68), but the latter one didn't go along with my ``Netscape''
  setup - it gave me some really huge font in the menubar. The xrus
  package doesn't have this problem.

  I would suggest you to download and try both of them. Pick up the one
  which you'll like more. Also, I'm going to creat RPM packages soon for
  both collections and download them to ftp.redhat.com.

  There are also older stuff, for example the vakufonts package, created
  by Serge Vakulenko, which was the base for the one in the X
  distribution. There are also a number of others. The important point
  is that the fonts' names in the old collection were not strictly
  conforming to the standard. The latter is fine in general, but
  sometimes it may cause various weird errors. For example, I had a bad
  experience with Maple V for Linux, which crashed mysteriously with the
  vakufonts package, but ran smoothly with the "standard" ones.

  So, let's start with the fonts:


  1. Download the appropriate fonts collection. The package for XFree86
     may be found at any FTP site, containing the X distribution, for
     example, directly from the XFree86 FTP site. The xrus package may
     be found on ftp.kiae.su

  2. Now when you have the fonts, you create some directory for them. It
     is generally a bad idea to put new fonts to the already existing
     font directory. So, place them, to, say,
     /usr/lib/X11/fonts/cyrillic for a system-wide setup, or just create
     a private directory for personal use.

  3. If the new fonts are in BDF format (*.bdf files), you have to
     compile them. For each font do:


     bdftopcf -o <font>.pcf <font>.bdf



  If your server supports compressed fonts, do it, using the compress
  program:


  compress *.pcf



  Also, if you do want to put the new fonts to an already existing font
  directory. you have to concatenate the old and the new files named
  fonts.alias in the case both of them exist.

  4. Each font directory in the X must contain a list of fonts in it.
     This list is stored in the file fonts.dir. You don't have to create
     this list manually. Instead, do:


     cd <new font directory>
     mkfontdir .



  5. Now you have to make this font directory known to the X server.
     Here, you have a number of options:


  o  System-wide setup for XFree86. If you are running this version of
     X, then append the new directory to the list of directories in the
     file XF86Config. To find the location of this file, see output of
     startx. Also, see XF86Config(4/5) for details.

  o  System-wide setup through xinit. Add the new directory to the xinit
     startup file. See xinit(1x) and the next option for details.


  o  Personal setup. You have a special start-up file for the X -
     ~/.xinitrc (or ~/.Xclients, or ~/.xsession for the RedHat users).
     Add the following commands to it:



     xset +fp <new font directory>
     xset fp rehash



  It is important to note that '+fp' means that the new fonts will be
  added to the head of the font path list. That is, if an application
  requests say a fixed font, it'll be given the one with Cyrillic char-
  acters, which is definitely what we are trying to achieve.

  There are problems, though. The fixed font in the cyrillic fonts dis-
  tribution doesn't have it's bold and italic counterparts. My font of
  choice is 6x13, so, since it also lacks bold and italic typefaces, I
  cannot use Emacs/XEmacs faces in their full glory. Hopefully somebody
  will ultimately create those fonts and the situation will change.

  6. Now restart your X. If you have done everything right, the tests in
     the beginning of the section will be successful. Also, play with
     xfontsel(1x) to make sure you are able to select the cyrillic
     fonts.

  In order to make the X clients use the Cyrillic fonts, you have to set
  up the appropriate X resources. For example, I make the russian font
  the default one in my ~/.Xdefaults:


  *font:         6x13



  Since my cyrillic fonts are first in the font path (see output of

  This just a simple case. If you want to set the appropriate part of
  the X client to a cyrillic font, you have to figure out the name of
  the resource (eg. using editres(1x)) and to specify it either in the
  resource database, or in the command line. Here go some examples:


  $ xterm -font '-cronyx-*-bold-*-*-*-19-*-*-*-*-*-*-*'



  ...will run xterm with some ugly font; and


  $ xfontsel -xrm '*quitButton.font: -*-times-*-*-*-*-13-*-*-*-*-*-koi8-*'



  ...will set a Cyrillic Times font for the Quit button in xfontsel.



  3.2.2.  The input translation


  In the newest X releases (X11R61 and higher) there are two "standard"
  input methods: the original one, working through the xmodmap utility,
  and the new one called Xkb (X KeyBoard). The very first thing you have
  to do is to disable the Xkb method! Don't get charmed by it's ability
  to set up a "russian keyboard". It looks like this method is using the
  Cyrillic keysyms defined in keysymdef.h. This file defines keysyms for
  many languages. The only problem is that those definitions have
  nothing to do with the extended ASCII codeset - the one most programs
  are only able to operate with! I hardly know any programs being able
  to grok the keysymdef.h keysyms, different from 8-bit ASCII. However
  our goal is to get the KOI8-R support to work.

  To disable the Xkb support, browse through the Keyboard section of
  your XF86Config file and comment all lines starting with Xkb (case
  doesn't matter). Instead, put the following line:


  XkbDisable



  The xmodmap program.allows customization of codes emitted by various
  characters and their combinations. It sets the things up based on the
  file containing the translation table.

  In the previous versions of this document I used to describe the
  xmodmap-based setup in a great detail. This proved to be almost
  useless. The Xmodmap-based input translation method is well known as
  being it is non-portable, inflexible, and incomplete.  Your
  configuration may work with one XFree version and fail with a
  different one. Even worse, sometimes things differ accross different
  servers in the same distribution.

  I strongly suggest you not to play with this xmodmap, at least for
  now. Apart from headache and disappointment you'll gain nothing.
  Instead, I recommend installing the xruskb package, which allows you
  to configure most of the input translation parameters without having
  to know about xmodmap. Again, the RedHat Linux users are free to
  download and install an RPM package.



  3.3.  First steps - Cyrillic in shells



  3.4.  bash


  Three variables should be set on order to make bash understand the
  8-bit characters. The best place is ~/.inputrc file. The following
  should be set:


  set meta-flag on
  set convert-meta off
  set output-meta on





  3.5.  csh/tcsh


  The following should be set in .cshrc:




  setenv LC_CTYPE iso_8859_5
  stty pass8



  If you don't have the POSIX stty (impossible for Linux), then replace
  the last call to the following:


  stty -istrip cs8





  3.6.  ksh


  As for the public domain ksh implementation - pdksh 5.1.3, you can
  input 8 bit characters only in vi input mode. Use:


  set -o vi





  3.7.  less


  So far, less doesn't support the KOI8-R character set, but the
  following environment variable will do the job:


  LESSCHARSET=latin1





  3.8.  mc (The Midnight Commander)


  To display Cyrillic text correctly, select the full 8 bits item in the
  Options/Display menu.

  If your problem is the ugly windows' borders, consult the ``''
  section.

  As an off-topic, if you want to make mc use color in an Xterm window,
  set the variable COLORTERM:


  COLORTERM= ; export COLORTERM





  3.9.  rlogin


  Make sure that the shell on the destination site is properly set up.
  Then, if your rlogin doesn't work by default, use 'rlogin -8'.

  3.10.  zsh


  Use the same way as with csh (see section ``csh''). The startup files
  in this case are .zshrc or /etc/zshrc.



  4.  Editing text


  In this section I'll describe how to customize various text editors to
  work with Cyrillic text. This doesn't cover the word processors, which
  will be described later (see section ``'').



  4.1.  Emacs and XEmacs


  There are two version of the Emacs editor - GNU Emacs and XEmacs.
  While they provide more or less same functionality, some
  implementation details are significantly different. Cyrillic setup
  requires some low-level (in Emacs Lisp sense) tweaking, and it differs
  a bit for those two versions.

  NOTE: Apart from the setup described here, there is an alternative way
  to configure both versions of emacs - use MULE (MULtilanguage Emacs
  support). The latter way is fairly complicated and (to the best of my
  knowledge) rarely used, so I don't discuss it here.

  The minimal cyrillic support in GNU emacs (you don't have to do it for
  the XEmacs) is done by adding the following calls to one's .emacs
  (provided that the Cyrillic character set support is installed for
  console or X respectively):


  (standard-display-european t)

  (set-input-mode (car (current-input-mode))
     (nth 1 (current-input-mode))
     0)



  This allows the user to view and input documents in Russian.

  However, it isn't enough. Emacs doesn't know yet, that Cyrililic
  characters may constitute a word, let alon the upper/lower case
  conversion rules. In order to teach Emacs doing that, you have to
  modify the syntax and case tables of emacs:


  (require 'case-table)

  (let* ((ruc "\341\342\367\347\344\345\263\366\372\351\352\353\354\355\356\357\360\362\363\364\365\346\350\343\376\373\375\370\371\377\374\340\361")
         (rlc "\301\302\327\307\304\305\243\326\332\311\312\313\314\315\316\317\320\322\323\324\325\306\310\303\336\333\335\330\331\337\334\300\321")
         (i 0)
         (len (length ruc)))
    (while (< i len)
      (modify-syntax-entry (elt ruc i) "w  ")
      (modify-syntax-entry (elt rlc i) "w  ")
      (set-case-syntax-pair (elt ruc i) (elt rlc i) (standard-case-table))
      (setq i (+ i 1))))


  For this purpose I created a rusup.el file which does this, as well as
  a couple handy functions. You have to load it in your ~/.emacs.

  Finally, the russian.el <http://www.math.uga.edu/~valery/russian.el>
  package by Valery Alexeev (valery@math.uga.edu) allows the user to
  switch between cyrillic and regular input mode and to translate the
  contents of a buffer from one Cyrillic coding standard to another
  (which is especially useful while reading the texts imported from MS-
  DOS or Windows).



  4.2.  Using vi


  The vi editor (at least it's clone vim, available in most Linux
  distributions) is aware of 8-bit characters. It will allow you to
  enter cyrillic characters and will be able to recognize the word
  boundaries correctly. I don't know about the upper-/lower-case
  conversion rules, since I don't use vi much. If you know something
  about it, please inform me.



  4.3.  Editing text with joe


  Joe requires a special -asis option to recognize 8-bit characters. You
  may either specify this option at the command line, or to put it in
  ~/.joerc file (for personal use, or in /usr/lib/joerc for system-wide
  setup.

  If your program doesn't understand -asis option, you have to upgrade
  to the newer version.

  However, joe doesn't seem to understand the cyrillic words' boundaries
  correctly. I assume, that it applies both to the case conversion
  rules.



  4.4.  Spell-checking Russian


  The program I use to spell-check text is the GNU ispell. It is very
  flexible and extensible, so it is possible to use it to spell-check
  text in languages, other than English, by adding new spell
  dictionaries.

  Constantine Knizhnik has created a very good Russian dictionary for
  ispell. You may find it at his homepage. The distribution includes a
  handy incremental spelling script for emacs.

  Ideally, if you already have an ispell properly installed, you have to
  just step into the newly-created directory and generate the
  dictionary, using the commands provided in the Makefile. However,
  chances are quite high, that you'll see a lot of complaints about the
  ispell's unawareness of the 8-bit data. This is because in most
  distributions, ispell is compiled without 8-bit data support. In this
  case, you cannot avoid recompiling the ispell package.

  Again, RedHat users will be delighted to know that I've rebuilt the
  ispell package with both Russian and German dictionaries. As usual,
  you may grab it from the RedHat FTP site.


  Once you have everything installed, you may invoke Russian spell-
  check, by supplying '-d russian' option to ispell.

  Now, if you use Emacs, you may want to add a menu item for a russian
  dictionary. I sent a proposed menu entry to the ispell.el maintainer
  and he kindly agreed to include it in the the next public release of
  the file. Meanwhile, you may do it by adding the following code in
  your ~/.emacs (or in /usr/share/emacs/site-lisp/site-start.el for a
  system-wide setup):


  (setq ispell-dictionary-alist
    (append ispell-dictionary-alist
            '(("russian"
               "[\341\342\367\347\344\345\263\366\372\351\352\353\354\355\356\357\360\362\363\364\365\346\350\343\376\373\375\370\371\377\374\340\361\301\302\327\307\304\305\243\326\332\311\312\313\314\315\316\317\320\322\323\324\325\306\310\303\336\333\335\330\331\337\334\300\321]"
               "[^\341\342\367\347\344\345\263\366\372\351\352\353\354\355\356\357\360\362\363\364\365\346\350\343\376\373\375\370\371\377\374\340\361\301\302\327\307\304\305\243\326\332\311\312\313\314\315\316\317\320\322\323\324\325\306\310\303\336\333\335\330\331\337\334\300\321]"
               "[']" t ("-C" "-d" "russian") "~latin1"))))

  (define-key-after ispell-menu-map [ispell-select-russian]
    '("Select Russian (KOI-8)" . (lambda ()
                                   (interactive)
                                   (ispell-change-dictionary "russian")))
    'british)



  Unfortunately, it won't work for the XEmacs. I'll try to solve this
  problem later.



  5.  Using Cyrillic with mail and news


  Setting up your mail and news software to recognize Cyrillic text is
  not very difficult, although you have to possess some knowledge of
  principles, mail and news work by.

  Internet electronic mail software generally consists of two parts: MUA
  (Mail User Agent) and MTA (Mail Transfer Agent). MUA is the program
  you use to read, compose, and send mail. However, MUA doesn't transfer
  mail messages by itself. Instead, it calls the MTA, which is
  reponsible to send message using an appropriate protocol to the
  appropriate direction. For example, your MUA may be Pine and MTA -
  qmail.

  Until quite recently, both MTA and MUA weren't 8-bit clean by default.
  Therefore, whenever you sent your message from say America to Russia,
  you were never sure, that some intermediate MTA won't strip the 8th
  bit from each character of your message. Therefore, a set of protocols
  was developed, which allowed encoding various kinds of data using only
  printable characters from 7-bit ASCII. This family of protocols is
  called MIME (MultimedIa Mail Encoding).

  Since MIME is usually pre-configured to reasonable defaults, we won't
  describe it here. We will talk more about MIME when we provide a
  backward compatibility with other Cyrillic encodings (section ``'').

  Meanwhile, we start MUA setup, because it is usually up to an end-
  user. Then, we will describe the basic priciples of the MTA
  configuration for Cyrillic.





  5.1.  Setting up Mail User Agents




  5.1.1.  Emacs-based mail readers


  Basically, you don't need any special setup for Emacs-based readers,
  geivedn, that you've already configured the emacs itself (see section
  ``'').



  5.1.2.  pine


  Set the following directive in ~/.pinerc for personal configuration,
  or in /usr/lib/pine.conf for a global one:


  character-set=ISO-8859-5





  5.2.  Configuring your MTA


  There are a number of MTAs available now. These include sendmail,
  qmail, smail, exim, and others.



  5.2.1.  sendmail


  So far, sendmail is much more popular than other MTAs, because it's
  long history and widespread use. Personally, I hate this program - it
  is a perfect example of a completely moronic design and even it's
  "improvements" with the passion of time show, that this approach is
  not going to cease. Any system administrator shudders, when he hears
  the ominous "sendmail.cf" name...

  As of now, sendmail doesn't strip the 8th bit anymore. However, it may
  encode the 8-bit data using a special base64 encoding. Although most
  MUAs are supposed to recognize it and decode it back to a regular
  data, you may want to start with sending raw 8-bit text to make sure
  everything works.

  As of version 8, sendmail handles 8-bit data correctly by default. If
  it doesn't do it for you, check the EightBitMode option and option 7
  given to mailers in your /etc/sendmail.cf. See "Sendmail. Operation
  and Installation Guide" for details.



  5.2.2.  Other MTAs


  I don't know much about other MTAs. If you know something, which may
  be important for Cyrillic setup, please inform me.



  6.  Browsing the Cyrillic Web


  Unlike e-mail and news, there is no definitive standard for Cyrillic
  encoding for the Web. This is primarily because Microsoft offers Web
  authoring tools, which only allow cp1251 codeset for Cyrillic,
  completely ignoring the fact that any other standards may already
  exist.

  The setup described here is very basic. It will allow you to view
  pages in the KOI8-R codeset. If the situation improves, I'll add more
  information.



  6.1.  lynx


  As of version 2.6, you may select the appropriate encoding for the
  display Character set option.



  6.2.  Netscape navigator


  Make sure you are using Netscape version higher than 3. If your
  Netscape is older, download a new one from www.netscape.com.



  6.2.1.  Basic setup


  To be able to see Cyrillic text in most parts of the HTML document, do
  the following:


  o  In menu Options/Document Encoding select Cyrillic(KOI-8).

  o  In menu Options/General Preferences/Fonts select Cyrillic (KOI-8)
     encoding, Times(Cronyx) as a proportional font and Courier(Cronyx)
     as a fixed one.

  o  save options.

  NOTE: This setup will work with most parts of the document. However,
  you won't be able to display Cyrillic text in the window header, menus
  and some controls. Attempts to fix it follows.



  6.2.2.  Cyrillic text in frames and input areas


  To fix this, it is usually enough to:


  1. Copy the Netscape properties database (usually Netscape.ad) to
     ~/Netscape.

  2. In the latter file, set the following property:


     *documentFonts.charset*iso8859-1:           koi8-r

  This will force all frame and input elements to use the fonts with
  koi8-r encoding instead of the default ones, therefore you have to
  make sure you have installed such fonts (see section ``'').

  The bad news about the trick above is that if you load a document
  which is supposed to be displayed in iso-8859-1 fonts, it will be
  displayed using the koi8 fonts instead. Sometimes such documents will
  look worse.



  6.2.3.  Advanced setup


  Andrew A. Chernov is the one, who knows more than others about KOI-8
  in general and netscape in particular. Visit his excellent KOI-8 page
  and download a patch for Netscape resource file, making Netscape speak
  Russian as much as it is able to.



  7.  Cyrillic wordprocessing




  7.1.  TeX-based environments


  In this section I'll describe several ways to make TeX and LaTeX
  typeset Cyrillic texts. There are several ways, which differ in setup
  sophistication and usage convenience. For example, one possibility is
  to start without any preliminary setup and use the Washington AMSTeX
  Cyrillic fonts. On the other hand, you may install a LaTeX package,
  providing a very high degree of Cyrillic setup. I have an experience
  with two such packages. One is the cmcyralt package by Vadim V.
  Zhytnikov (vvzhy@phy.ncu.edu.tw) and Alexander Harin
  (harin@lourie.und.ac.za), and the other one is the LH package by the
  CyrTUG group with styles and hyphenation for LaTeX2e by Sergei O.
  Naoumov (serge@astro.unc.edu). I'll describe both.

  Note, that there are two versions of LaTeX available - 2.09 is the old
  one, while 2e is a new pre-3.0 release. If you are using LaTeX 2.09,
  then switch quickly to the 2e. The latter retains compatibility with
  the old one, but has much more features. Hopefully, version 3 will be
  released soon. I describe a LaTeX 2e setup.

  Also, both of these packages require the Cyrillic text to be typeset
  using the Alt codeset, not KOI8-R! This is caused by historical
  reasons, since the creators of these packages used to work with EmTeX
  - the MS-DOG version of TeX (they didn't know about Linux yet :-).
  Switching to the KOI8-R requires some effort and is being expected to
  be done soon. So far, use some utility to convert your russian text
  from KOI8-R to Alt. See section ``''.



  7.1.1.  Using the Washington Cyrillic


  This package was created for the American Mathematic Society to
  provide documents with Russian references. Therefore, the authors were
  not very careful and the fonts look quite clumsy. This package is
  usually referred to as a "really bad cyrillic package for TeX".


  Nevertheless, we'll discuss it, because it is very easy to use and
  doesn't require any setup - this collection is supplied with most of
  TeX distributions.

  Of course, you won't be able to use such luxury as automatic
  hyphenation, but anyway...

  1. Prepend your document with the following directives:


  \input cyracc.def
  \font\tencyr=wncyr10
  \def\cyr{\tencyr\cyracc}



  2. Now to type a cyrillic letter, you enter


  \cyr



  and use a corresponding latin letter or a TeX command. Thus, the lower
  case of the Russian alphabet is expressed by the following codes:


  a b v g d e \"e zh z i {\u i} k l m n o p r s t u f kh c ch sh shch
  {\cprime} y {\cdprime} \`e yu ya



  It is extremely inconvenient to convert your Russian texts to such
  encoding, but you can automate the process. The translit program
  (section ``'') supports a TeX output option.



  7.1.2.  KOI-8 package for teTeX


  There is some new teTeX-rus package. It is reported to support KOI-8
  character set and have all basic stuff required for TeX and LaTeX. I
  personally haven't tried it yes, although I heard about it's
  successfull usage.

  NOTE: This package requires you to reconfigure and rebuild some parts
  of your teTeX package (for example the precompiled LaTeX macros).
  Unless you know what you are doing, you shouldn't try it without
  necessary care. Otherwise, you may be better off by borrowing the
  precompiled parts fron somebody on the net



  7.1.3.  Using the cmcyralt package for LaTeX


  The cmcyralt package can be found on any CTAN (Comprehensive TeX
  Archive Network) site like ftp.dante.de. You should obtain two pieces:
  the fonts collection from fonts/cmcyralt and the styles and
  hyphenation rules from macros/latex/contrib/others/cmcyralt.

  Note: Make sure you have the Sauter package installed, since cmcyralt
  requires some fonts from it. You can get this package from CTAN site
  as well.

  Now you should do the following:


  1. Put the new fonts to the TeX fonts tree. On my system (Slackware
     2.2) I created a cmcyralt directory in the
     /usr/lib/texmf/fonts/cm/. Create the src, tfm, and vf
     subdirectories in it. Put there .mf, .tfm, and vf files
     respectively.

  2. Put the font driver files (*.fd) from the styles archive to the
     appropriate place (in my case it was /usr/lib/texmf/tex/latex/fd).

  3. Put the style files (*.sty) to the appropriate LaTeX styles
     directory (in my case /usr/lib/texmf/tex/latex/sty).

  Now the hyphenation setup. This requires to remake the LaTeX base
  file.


  1. The file hyphen.cfg contains the directives for both English and
     Russian hyphenation. Extract the one for Russian and place it to
     the LaTeX hyphenation config file lthyphen.ltx. In my case, that
     file was in /usr/lib/texmf/tex/latex/latex-base.

  2. Put the rhyphen.tex to the same directory. It is needed for making
     the new base file. Later, you can remove it.

  3. Do 'make' in that directory. Don't for get to make a link from
     Makefile to Makefile.unx. During the make process check the output.
     There should be a message:


     Loading hyphenation patterns for Russian.



  If everything goes OK, you will get the new latex.fmt in that direc-
  tory. Put it to the appropriate place, where the previous one was
  (like /usr/lib/texmf/ini/). Don't forget to save the previous one!.

  This is it. The installation is complete. Try processing the examples
  found in the styles archive. If you are to create the PostScript files
  without any problems, then everything is OK. Now, to use Cyrillic in
  LaTeX, prepend your document with the following directive:


  \usepackage{cmcyralt}



  For more details, see the README file in the cmcyralt styles archive.

  Note: if you do have problems with the examples, provided you have
  installed the things right, then probably your TeX system hasn't been
  installed correctly. For example, during my first try, every attempt
  to create the .pk files for the russian fonts failed (MakeTeXPK
  stage). A substantial investigation discovered some implicit conflict
  between the localfont and ljfour METAFONT configurations. It used to
  work before, but kept crashing after the cmcyralt installation.
  Contact your local TeX guru - TeX is very (sometimes too much)
  complicated to reconfigure it without any prior knowledge.





  7.1.4.  Using the CyrTUG package


  You can obtain the CyrTUG package from the SunSite archive. Get the
  files CyrTUGfonts.tar.gz, CyrTUGmacro.tar.gz, and hyphen.tar.Z.

  The process of installation doesn't differ from too much the previous
  one.





  7.2.  The StarOffice suite


  Youri Kovalenko () has compiled a concise summary on StarOffice
  russification. It is located at .  I never had a chance to try it, so
  I cannot say anything about it's correctness.

  Another source of information on the subject is compiled by Eugene
  Demidov () and is located at .



  8.  Printing and PostScript




  8.1.  Text to PostScript conversion


  Sometimes you have just a plain ASCII KOI8-R text and you want to
  print it just to get it on the paper. One of the easiest ways to
  achieve that is to use special programs converting text to PostScript.

  There are a number of programs doing such conversion. I personally
  prefer a2ps. Originally developed as a simple text-to-PostScript
  converter it became a big and highly configurable program with many
  options and allows you to manage various page layouts, syntax
  highlighting etc. Another tool (now available as a part of the GNU
  project) is enscript.



  8.1.1.  An a2ps converter


  A text to PostScript converter has been around for a while and is one
  of the most versatile printing tools. The author proved to be very
  open to suggestions, so since the release 4.9.8 a2ps supports Cyrillic
  right off-the-shelf. All you need is a PostScript printer.

  The command I use is:


  a2ps -X koi8r --print-anyway  <file>








  8.1.2.  The GNU enscript


  The GNU enscript program is also designed for converting text to
  PostScript and it also has a non-ASCII codeset support. It doesn't
  have Cyrillic PostScript fonts, but it is very easy to get them, as
  will be explained below (thanks to Michael Van Canneyt):


  1. Install the newest enscript. As of now, the most recent release is
     1.5. You may either get the one from the GNU FTP archive, or take
     an RPM package from the Redhat site.

  2. Now, if you are a lucky RedHat Linux user, download and install
     Cyrillic Textbook font
     <ftp://ftp.redhat.com/pub/contrib/i386/enscript-fonts-
     koi8-1.0-1.i386.rpm>.

  3. If you don't use RPM, download a file textbook.tar.gz from the
     Cyrillic Software collection on sunsite.unc.edu
     <ftp://sunsite.unc.edu/pub/academic/russian-studies/Software/>.
     Extract it to a directory, where enscript fonts are located
     (usually /usr/share/enscript). Now change to that directory and run
     the following command:


         mkafmmap *.afm



  4. The setup is finished. Try to print some text in KOI8-R Cyrillic
     with the following command:


         enscript --font=Textbook8 --encoding=koi8 some.file



  If you want a really quick and dirty solution and you don't care about
  the output quality and all you need is just Cyrillic on the paper, try
  the rtxt2ps package. It is a very simple no-frills text-to-PostScript
  conversion program. The output quality is not very good (or, to be
  honest, just bad) but it does it's job.



  8.2.  Text to TeX conversion


  If all you need is just to print an ASCII text without any additional
  word processing, you may try to use some programs, which would convert
  your Cyrillic text to a ready-to-process TeX file. One of the best
  programs for such purposes is translit (see section ``''). In this
  case, you don't even have to bother about installing the Cyrillic
  fonts for TeX, since translit uses a Washington Cyrillic package,
  which is included in most TeX distributions (or am I wrong?)



  9.  Cyrillic in PostScript


  Experts say PostScript is easy. I cannot judge - I've got too many
  things to learn to spare some time to learn PostScript. So I'll try to
  use my sad experience with it. I'll appreciate any feedback from you
  guys who know more on the subject than I do (approx. 99% of the Earth
  population).

  Basically, in order to print a Cyrillic text using PostScript, you
  have to make sure about the following things:


  o  Cyrillic font is loaded or included in the document.

  o  Cyrillic text is included in the document.

  o  Cyrillic text uses the appropriate character codes which correspond
     to the font's requirements.

  o  An appropriate font is selected in order to print Cyrillic text.

  There is no solution general enough to be recommended as an ultimate
  treatment. I'll try to outline various ways to cope with different
  problems related to the subject.

  One way to address Cyrillic setup problems generally enough is to use
  Ghostscript. Ghostscript (or just gs in the newspeak) is a free (well
  quasi-free) PostScript interpreter. It has many advantages; among
  them:


  o  Ability to run on many platforms (various Unices, Windows etc)

  o  Support for a wide number of non-PostScript printers

  o  Good degree of configurability

  What is important in our particular case, is that once Ghostscript is
  set up, we can do all printing through it, thus eliminating extra
  setup for other PostScript devices (for example HP LaserJet IV)



  9.1.  Adding Cyrillic fonts to Ghostscript


  This is important, since you probably don't want to put a
  responsibility to other programs to insert Cyrillic fonts in the
  PostScript output. Instead, you add them to gs and just make the
  programs generate Cyrillic output compatible with the fonts.

  To add a new font (in pfa or pfb form) in gs, you have to:


  1. Put it in the gs fonts directory (ie.  /usr/lib/ghostscript/fonts).

  2. Add the appropriate names and aliases for the font in the Fontmap
     file in the gs directory.

  Recently a decent set of Cyrillic fonts for GhostScript appeared.  It
  is located in ftp.kapella.gpi.ru. This one even has a necessary part
  to add to the Fontmap file. You have to download the contents of the
  /pub/cyrillic/psfonts directory. The README file describes the
  necessary details.



  10.  Print setup


  Printing is always tricky. There are different printers from different
  vendors with different facilities. Even for a native printing there is
  no uniform solution (this applies not only to UNIX, but to other
  operating systems as well.

  Printers have different control languages and often they have very
  different views on foreign language support.  The good news is that on
  control language seems to be recognized as a de-facto standard for
  print job description - it is a PostScript language developed by Adobe
  Corporation.

  Another problem is a variety of requirements to the print services.
  For example, sometimes you want just to print a piece if C program,
  containing comments in Russian, so you don't need any pretty-printing
  - just a raw ASCII output in a single font. Another time, when you
  design a postcard for your girlfriend, you'll probably need to typeset
  some document with different fonts etc. This will definitely require
  more effort to setup Cyrillic support.

  To accomplish the former task you just have to make your printer
  understand one Cyrillic font and (maybe) install some filter program
  to generate data in appropriate format. To accomplish the latter one,
  you have to teach your printer different fonts and have a special
  software.

  There is also something in the middle, when you get a program which
  knows how to generate both the fonts and the appropriate printer
  input, so you can say do some aource code pretty-printing without
  sophisticated word processing systems.

  All these options will be more or less covered below.



  10.1.  Pre-loading Cyrillic fonts into a non-PostScript printer


  If you have a good old dot matrix printer and all you need is to print
  a raw KOI8-R text, try the following:


  1. Find a proper KOI8-R font for your printer. Check out the MS-DOSish
     stuff on the Internet (for example the SimTel archive
     <ftp://ftp.simtel.net>).

  2. Learn from the manual, how to load such font into your printer and,
     probably, write a simple program doing that.

  3. Run this program from the appropriate rc file at a boot time.

  Thus, having Cyrillic characters in the upper part of the printer's
  character set will allow you to print you texts in Russian without any
  hussle.

  Alternatively to the KOI8-R fonts you may try to use the Alt font.
  There are two reasons for that:


  o  It may be probably much easier to find an Alt font, since those
     were very widespread in the MS-DOS culture.

  o  Having a proper Alt font will allow you to print pseudo-graphic
     characters as well.

  However in this case, you'll have to convert your texts from KOI8-R to
  Alt before sending them to a printer. This is quite easy, since there
  are a lot of programs doing that (see ``translit'' for example), so
  you just have to call such program properly in the if field in
  /etc/printcap file. For example, with the translit program you may
  specify:



  if=/usr/bin/translit -t koi8-alt.rus



  See printcap(5) for details.



  10.2.  Printing with different fonts


  One great way to cope with different printers and fonts is to use TeX
  (see section ``''). TeX drivers handle all details, so once you make
  TeX understand Cyrillic fonts, you are done.

  Another possibility is to use PostScript. I decided to devote an
  entire chapter ``'' to the subject, since it is not simple.

  Finally, there are other word processors, which have printer drivers.
  I never tried anything apart from TeX, so I cannot suggest anything.



  11.  Localization and Internationalization


  So far, I described how to make various programs understand Cyrillic
  text. Basically, each program required it's own method, very different
  from the others. Moreover, some programs had incomplete support of
  languages other than English. Not to mention their inability to
  interact using user's mother tongue instead of English.

  The problems outlined above are very pressing, since software is
  rarely developed for home market only. Therefore, rewriting
  substantial parts of software each time the new international market
  is approached is very ineffective; and making each program implement
  it's own proprietary solution for handling different languages is not
  a great idea in a long term either.

  Therefore, a need for standardization arises. And the standard shows
  up.

  Everything related to the problems above is divided by two basic
  concepts: localization and internationalization. By localization we
  mean making programs able to handle different language conventions for
  different countries. Let me give an example. The way date is printed
  in the United States is MM/DD/YY. In Russia however, the most popular
  format is DD.MM.YY. Another issues include time representation,
  printing numbers and currency representation format. Apart from it,
  one of the most important aspect of localization is defining the
  appropriate character classes, that is, defining which characters in
  the character set are language units (letters) and how they are
  ordered. On the other hand, localization doesn't deal with fonts.

  Internationalization (or i18n for brevity) is supposed to solve the
  problems related to the ability of the program interact with the user
  in his native language.

  Both of the concepts above had to be implemented in a standard, giving
  programmers a consistent way of making the programs aware of national
  environments.
  Althogh the standard hasn't been finished yet, many parts actually
  have; so they can be used without much of a problem.

  I am going to outline the general scheme of making the programs use
  the features above in a standard way. Since this deserves a separate
  document, I'll just try to give a very basic description and pointers
  to more thorough sources.



  11.1.  Locale


  One of the main concept of the localization is a locale. By locale is
  meant a set of conventions specific to a certain language in a certain
  country. It is usually wrong to say that locale is just country-
  specific. For example, in Canada two locales can be defined -
  Canada/English language and Canada/French language. Moreover,
  Canada/English is not equivalent to UK/English or US/English, just as
  Canada/French is not equivalent to France/French or
  Switzerland/French.



  11.1.1.  How to use locale


  Each locale is a special database, defining at least the following
  rules:


  1. character classification and conversion

  2. monetary values representation

  3. number representation (ie. the decimal character)

  4. date/time formatting


  In RedHat 4.1, which I am using there are actually two locale
  databases: one for the C library (libc) and one for the X libraries.
  In the ideal case there should be only one locale database for
  everything.

  To change your default locale, it is usually enough to set the LANG
  environment variable. For example, in sh:


  LANG=ru_RU
  export LANG



  Sometimes, you may want to change only one aspect of the locale
  without affecting the others. For example, you may decide (God knows
  why) to stick with ru_RU locale, but print numbers according to the
  standard POSIX one. For such cases, there is a set of environment
  variables, which you can you to configure specific parts for the
  current locale. In the last exaple it would be:


  LANG=ru_RU
  LC_NUMERIC=POSIX
  export LANG LC_NUMERIC

  For the full description of those variables, see locale(7).

  Now let's be more Linux-specific. Unfortunately, Linux libc version
  5.3.12, supplied with RedHat 4.1, doesn't have a russian locale. In
  this case one must be downloaded from the Internet (I don't know the
  exact address, however).

  To check, locale for which languages you have, run 'locale -a'. It
  will list all locale databases, available to libc.

  Fortunately, Linux community is rapidly moving to the new GNU libc
  (glibc version 2, which is much more POSIX-compliant and has a proper
  russian locale. Next "stable" RedHat system will already use glibc.

  As for the X libraries, they have their own locale database. In the
  version I am using (XFree86 3.3), there already is a russian locale
  database. I am not sure about the previous versions. In any case, you
  may check it by looking into usr/lib/X11/locale/ (on most systems). In
  my case, there already are subdirectories named koi8-r and even
  iso8859-5.



  11.1.2.  Locale-aware programming


  With locale, program don't have to implement explicitly various
  character conversion and comparison rules, described above. Instead,
  they use special API which make use of the rules defined by locale.
  Also, it is not necessary for program to use the same locale for all
  rules - it is possible to handle different rules using different
  locales (although such technique should be strongly discouraged).

  From the setlocale(3) manual page:


       A  program  may be made portable to all locales by calling
       setlocale(LC_ALL, "" ) after program   initialization,  by
       using  the  values  returned  from a localeconv() call for
       locale - dependent information and by using  strcoll()  or
       strxfrm() to compare strings.


  SunSoft, for example, defines 5 levels of program localization:


  1. 8-bit clean software. That is, the program calls setlocale(), it
     doesn't make any assumptions about the 8th bit of each character,
     it users functions from ctype.h and limits from limits.h, and it
     takes care about signed/unsigned issues.

     It is very important not to do any assumption about the character
     set nature and ordering. The following programming practices must
     be avoided:


         if (c >= 'A' && c <= 'Z') {
             ...



  Instead, macros from the ctype.h header file are locale-aware and
  should be used in all such occasions.

  2. Formats, sorting methods, paper sizes. The program uses strcoll()
     and strxfrm() instead of strcmp() for strings, it uses time(),
     localtime(), and strftime()/ for time services, and finally, it
     uses localeconv() for a proper numbers and currency representation.

  3. Visible text in message catalogs. The program must isolate all
     visible text in special message catalogs. Those map strings in
     English to their translation to other languages. Selection of
     messages in an appropriate for a particular environment language is
     done in a way which is completely transparent for both the program
     and it's user. To make use of those facilities, the program must
     call gettext() (Sun/POSIX standard), or catgets() (X/Open
     standard). For more information on that see section ``''.


  4. EUC/Unicode support. At this level, the program doesn't use the
     char type. Instead it uses wchar_t, which defines entities big
     enough to contain Unicode characters. ANSI C defines this data type
     and an appropriate API.


  For a more detaled explanation of locale, see, for example (``'') or
  (``'').



  11.2.  Internationalization


  While localization describes, how to adapt a program to a foreign
  environment, internationalization (or i18n for brevity) details the
  ways to make program communicate with a non-English speaking user.

  Before, that was done by developing some abstraction of the messages
  to output from the program's code. Now, such mechanism is (more or
  less) standardized. And, of course, there are free implementations of
  it!

  The GNU project has finally adopted the way of making the
  internationalized applications. Ulrich Drepper (drepper@ipd.info.uni-
  karlsruhe.de) developed a package gettext. This package is available
  at all GNU sites like prep.ai.mit.edu. It allows you to develop
  programs in the way that you can easily make them support more
  languages. I don't intend to describe the programming techniques,
  especially because the gettext package is delivered with excellent
  manual.

  Request for collaboration: If you want to learn the gettext package
  and to contribute to the GNU project simultaneously; or even if you
  just want to contribute, then you can do it! GNU goes international,
  so all the utilities are being made locale-aware. The problem is to
  translate the messages from English to Russian (and other languages if
  you'd like). Basically, what one has to do is to get the special .po
  file consisting of the English messages for a certain utility and to
  append each message with it's equivalent in Russian. Ultimately, this
  will make the system speak Russian if the user wants it to! For more
  details and further directions contact Ulrich Drepper
  (drepper@ipd.info.uni-karlsruhe.de).




  12.  Staying compatible


  Being standard is not the only issue. To be really nice, one has to
  provide the backward compatibility. In our case, this means that the
  configuration should be tolerant to the data created using non-
  standard character sets - that is the Alt (cp866) and cp1251 ones.
  Also, we should be able to run Cyrillic programs for MS-DOS.

  In most cases (except for HTTP), it is enough to provide a timely
  conversion of data to KOI8-R. When we talk about raw unstructured
  data, it is quite trivial - see section ``Conversion Utilities''.

  Another issue is the structured data. This case is more tricky. I'll
  try to outline the basic roadmap of fixing it.



  12.1.  MIME-based data compatibility


  MIME is a standard for architecture-independent data representation.
  Originally developed for mail messages, it has now many more
  applications. MIME defines format, which is open to extensions and
  allows architecture-specific handling of data. For example, if I
  receive a mail message, containing a MIME object of the video/mpeg
  type (an encoded MPEG file), my mail reader will automatically decode
  it and start an MPEG player.

  Most UNIX programs, offering MIME capabilities, are based on the
  metamail package, which contains a set of utilities and data files to
  work with MIME objects. Several configuration files (/etc/mailcap for
  global usage and ~/.mailcap for personal setup) define rules for
  handling MIME object of various types.

  Thus, if you receive a proper MIME data stream, containing text in one
  of the obsolete character sets, you may define a MIME rule to convert
  such text to KOI8.

  Below a number of MIME rules are shown, which are supposed to handle
  plain text and richtext objects, using both of the obsolete codesets,
  discussed above. You may incorporate these rules into one of the MIME
  configuration files.

  Note, that these rules use the translit package to perform the actual
  conversion. For more information on that program and the conversion in
  general see section ``Conversion Utilities''.


  text/plain; translit -t cp1251-koi8.rus < %s; test=test \
      "`echo %{charset} | tr '[A-Z]' '[a-z]'`"  = cp1251; copiousoutput

  text/richtext; translit -t cp1251-koi8.rus < %s; test=test \
      "`echo %{charset} | tr '[A-Z]' '[a-z]'`"  = cp1251; copiousoutput

  text/plain; translit -t alt-koi8.rus < %s; test=test \
      "`echo %{charset} | tr '[A-Z]' '[a-z]'`"  = cp866; copiousoutput

  text/richtext; translit -t alt-koi8.rus < %s; test=test \
      "`echo %{charset} | tr '[A-Z]' '[a-z]'`"  = cp866; copiousoutput

  text/plain; translit -t alt-koi8.rus < %s; test=test \
      "`echo %{charset} | tr '[A-Z]' '[a-z]'`"  = alt; copiousoutput

  text/richtext; translit -t alt-koi8.rus < %s; test=test \
      "`echo %{charset} | tr '[A-Z]' '[a-z]'`"  = alt; copiousoutput



  Obviously enough, this will work for plain text data only. Binary
  files are supposed to handle the codeset issues themselves (at least
  their "parent" applications are). Therefore, if you receive a
  Microsoft Word document in the cp1251 character set, the duty of
  providing appropriate conversion capabilities lays upon an application
  you use to read that document (for example Microsoft Word, or Applix
  Words).

  Unfortunately, the real situation is not that ideal. Many application
  have their own idea on how to use MIME. Until recently Microsoft Mail
  software had a broken MIME engine. Also, the Netscape
  Navigator/Communicator mail client is notorious because of it's
  sending of mail messages, encoded in cp1251 with the charset=koi8-r
  field in the message header and vice versa.



  12.2.  Explicit character set conversion


  There are a lot of conversion routines for Cyrillic on the Internet.
  Each of them has it's own quirks and it's own degree of Cyrillic
  support.

  In my opinion tools must be standard. In this particular case the
  "standard" conversion tool is GNU recode. Unfortunately, the version,
  found on the official GNU site (3.4) doesn't support Cyrillic yet
  (only ISO-8859-5). I developed a set of conversion tables for KOI8-R,
  Alt, and cp1251 for recode and submitted them to the recode
  maintainer. He promised to provide Cyrillic support in the upcoming
  release. Once it happens, I'll rewrite this section to recommend GNU
  recode as the standard conversion engine for Cyrillic.

  Meanwhile, I would recommend a translit package. It supports many
  popular codesets and is even able to produce a *TeX files (see section
  ``'') from text in Russian. Also, RedHat users will enjoy an RPM
  package for translit.

  For other conversion routines, Look at SovInformBureau or
  ftp.funet.fi. You can even use the special mode for emacs (see section
  ``Emacs'').



  12.3.  Cyrillic in the DOS emulator


  This seems to be the only application, which may require Alt Cyrillic
  character set. The reason is that Alt is native to DOS and most of DOS
  programs dealing with Cyrillic are Alt-oriented.

  For the console version (dos) you just have to load a keyboard and
  screen driver. Most of DOS drivers will work fine. I personally use
  the rk driver by A. Strakhov, which works for both console and X
  versions of dosemu. Another choice is the r driver by V. Kurland
  (sorry for possible misspelling). It is perfectly customizable and
  supports many codesets, Alt and KOI8 among them. However it won't work
  for the X window (at least version 1.14 I'm using).

  Both drivers can be found on most Russian Internet sites, for example
  Kurchatov Institute FTP server <ftp://ftp.kiae.su/pub/cyrillic/msdos>.

  For the X version of dosemu you have to provide an appropriate X font
  as well. Alex Bogdanov sent me such font by e-mail. It is an original
  vga font from the dosemu distribution, modified for the Alt codeset.
  Unfortunately I don't know who is the creator of this font and where
  the official site is.


  To setup the font for dosemu you should


  o  Introduce this font to the X. This is described in ``X fonts
     setup''.

  o  Introduce this font to dosemu. If the font just replaces the
     original vga font, then it will be recognized by default.
     Otherwise, you have to describe it in /etc/dosemu.conf:


     # Font to use (without filename extensions). For example:
     X { updatefreq 8 title "MS DOS" icon_name "xdos" font "vga-alt"}



  Finally, you have to load a keyboard driver. Note, the you don't need
  a screen driver for the X window. Therefore, not all drivers will
  work. At least two will: rk by A. Strakhov, and cyrkeyb by Pete
  Kvitek.



  13.  Bibliography



  1. Andrey Chernov. KOI-8 <http://www.nagual.ru/~ache/koi8.html>. KOI-8
     information and setup.

  2. Ulrich Drepper. Internationalization in the GNU project
     <http://i44www.info.uni-karlsruhe.de/~drepper/conf96/paper.html>.
     Very thorough description of a GNU approach to i18n.

  3. Michael Karl Gschwind. Internationalization
     <http://www.vlsivie.tuwien.ac.at/mike/i18n.html>. Various resources
     on i18n.

  4. Sergei Naumov. Information on Cyrillic Software
     <http://sunsite.oit.unc.edu/sergei/Software/Software.html>.
     Cyrillic setup information.

  5. The Open Group Single UNIX specification <http://www.UNIX-
     systems.org/online.html>.

  6. RFC 1489 RFC 1489 <file://ds.internic.net/rfc/rfc1489.txt>

  7. Alec Voropay. Localization as it is
     <http://www.sensi.org/~alec/locale>. General locale usage in
     Russian.




  14.  Summary of the various useful resources


  a2ps homepage <http://www-inf.enst.fr/~demaille/a2ps.html>

  General Linux Information <http://www.linux.org>

  Collection of Cyrillic resources
  <ftp://ftp.ccl.net/pub/central\_eastern\_europe/russian>

  Cyrillic resources at KIAE <ftp://ftp.kiae.su/cyrillic/>

  Cyrillic resources at RELCOM <ftp://ftp.relcom.ru/cyrillic/>

  Cyrillic resources at FUNET
  <ftp://ftp.funet.fi/pub/culture/russian/comp/>

  Cronyx <http://www.cronyx.ru> - the creators of Cyrillic fonts for the
  X Window System.

  Cyrillic fonts for Ghostscript and StarOffice
  <ftp://ftp.kapella.gpi.ru/pub/cyrillic/psfonts>

  Cyrillic fonts for X
  <ftp://ftp.kiae.su/cyrillic/x11/fonts/xrus-2.1.1-src.tgz>

  Ghostscript <http://www.cs.wisc.edu/~ghost/index.html>

  GNU enscript <ftp://prep.ai.mit.edu/pub/gnu>

  relcom.fido.ru.linux newsgoup.

  relcom.fido.ru.unix newsgoup.

  Russian dictionary for GNU ispell <http://www.ispras.ru/~knizhnik>

  SovInformBureau <http://www.siber.com/sib/russify/>

  teTeX russification package <ftp://xray.sai.msu.su/pub/outgoing/teTeX-
  rus/>

  The kbd package for Linux
  <ftp://sunsite.unc.edu/pub/Linux/system/Keyboards/>

  The remap package for Emacs <ftp://ftp.iesd.auc.dk/>

  The rtxt2ps package <http://www.siber.com/sib/russify/converters/>

  The russian.el package for emacs
  <http://www.math.uga.edu/~valery/russian.el>

  The translit package
  <ftp://ftp.osc.edu/pub/russian/translit/translit.tar.Z>

  The xruskb package <ftp://ftp.relcom.ru/pub/x11/cyrillic/>

  Useful Cyrillic packages <ftp://sunsite.unc.edu/pub/academic/russian-
  studies/Software>

  X fonts collections <ftp://ftp.switch.ch/mirror/linux/X11/fonts/>

  XFree86 FTP site <http://www.xfree86.org>