ISO 8859-1 National Character Set FAQ

ISO 8859-1 National Character Set FAQ

Post by m.. » Sat, 21 Jan 1995 03:33:14



Archive-name: internationalization/iso-8859-1-charset
Posting-Frequency: monthly
Version: 2.6

                  ISO 8859-1  National Character Set FAQ

DISCLAIMER: THE AUTHOR MAKES NO WARRANTY OF ANY KIND WITH REGARD TO
THIS MATERIAL, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

Note: Most of this was tested on a Sun 10, running SunOS 4.1.* - other
systems might differ slightly

This FAQ discusses topics related to the use of ISO 8859-1 based 8 bit
character sets. It discusses how to use European (Latin American)
national character sets on UNIX-based systems and the Internet.

If youu need to use a character set other than ISO 8859-1, much of
what is described here will be of interest to you.  However, you will
need to find appropriate fonts for your character set (see section 17)
and input mechanisms adapted to you language.

1. Which coding should I use for accented characters?
Use the internationally standardized ISO-8859-1 character set to type
accented characters. This character set contains all characters
necessary to type (West) European languages. This encoding is also the
preferred encoding on the Internet.  ISO 8859-X character sets use the
characters 0xa0 through 0xff to represent national characters, while
the characters in the 0x20-0x7f range are those used in the US-ASCII
(ISO 646) character set.  Thus, ASCII text is a proper subset of all
ISO 8859-X character sets.  

The characters 0x80 through 0x9f are earmarked as extended control
chracters, and are not used for encoding characters.  These characters
are not currently used to specify anything.  A practical reason for
this is interoperability with 7 bit devices (or when the 8th bit gets
stripped by faulty software).  Devices would then interpret the character
as some control character and put the device in an undefined state.
(When the 8th bit gets stripped from the characters at 0xa0 to 0xff, a
wrong character is represented, but this cannot change the state of a
terminal or other device.)

This character set is also used by AmigaDOS, MS-Windows, VMS (DEC MCS
is practically equivalent to ISO 8859-1) and (practically all) UNIX
implementations.  MS-DOS normally uses a different character set and
is not compatible with this character set. (It can, however, be
translated to this format with various tools. See section 5.)

Footnote: Supposedly, IBM code page 819 is fully ISO 8859-1 compliant.

ISO 8859-1 supports the following languages:
Afrikaans, Basque, Catalan, Danish, Dutch, English, Faeroese, Finnish,
French, Galician, German, Icelandic, Irish, Italian, Norwegian,
Portuguese, Spanish and Swedish.

(It has been called to my attention that Albanian can be written with
ISO 8859-1 also.  However, from a standards point of view, ISO 8859-2
is the appropriate character set for Balkan countries.)

ISO 8859-1 is just one part of the ISO-8859 standard, which specifies
several character sets:
8859-1  Europe, Latin America
8859-2  Eastern Europe
8859-3  SE Europe/miscellaneous (Esperanto, Maltese, etc.)
8859-4  Scandinavia/Baltic (mostly covered by 8859-1 also)
8859-5  Cyrillic
8859-6  Arabic
8859-7  Greek
8859-8  Hebrew
8859-9  Latin5, same as 8859-1 except for Turkish instead of Icelandic
8859-10 Latin6, for Lappish/Nordic/Eskimo languages

Unicode is advantageous because one character set suffices to encode
all the world's languages, however very few programs (and even fewer
operating systems) support wide characters. Thus, only 8 bit wide
character sets (such as the ISO 8859-X) can be used with these
systems.  Unfortunately, some programmers still insist on using the
`spare' eigth bit for clever tricks, crippling these programs such
that they can process only US-ASCII characters.

Footnote: Some people have complained about missing characters,
          e.g. French users about a missing 'oe'.  Note that oe is
          not a character, but a ligature (a combination of two
          characters for typographical purposes).  Ligatures are not
          part of the ISO 8859-X standard.  (Although 'oe' used to
          be in the draft 8859-1 standard before it was unmasked as
          `mere' ligature.)

2. Getting your terminal to handle ISO characters.
Terminal drivers normally do not pass 8 bit characters. To enable
proper handling of ISO characters, add the following lines to your
.cshrc:
----------------------------------
tty -s
if ($status == 0) stty cs8 -istrip -parenb
----------------------------------
If you don't use csh, add equivalent code to your shell's start up
file.  

Note that it is necessary to check whether your standard I/O streams
are connected to a terminal. Only then should you reconfigure the
terminal driver.  Note that tty checks stdin, but stty changes stdout.
This is OK in normal code, but if the .cshrc is executed in a pipe,
you may get spurious warnings :-(

If you use the Bourne Shell or descendants (sh, ksh, bash,
zsh), use this code in your startup (e.g. .profile) file:
----------------------------------
tty -s
if [ $? = 0 ]; then
        stty cs8 -istrip -parenb >&0
fi
----------------------------------

Footnote: In the /bin/sh version, we redirect stdout to stdin, so both
tty and stty operate on stdin.  This resolves the problem discussed in
the /bin/csh script version.  A possible workaround is to use the
following code in .cshrc, which spawns a Bourne shell (/bin/sh) to
handle the redirection:
----------------------------------
tty -s
if ($status == 0) sh -c "stty cs8 -istrip -parenb >&0"
----------------------------------

3. Getting the locale setting right.
For the ctype macros (and by extension, applications you are running
on your system) to correctly identify accented characters, you
may have to set the ctype locale to an ISO 8859-1 conforming
configuration. On SunOS, this may be done by placing
------------------------------------
setenv LANG C
setenv LC_CTYPE iso_8859_1
------------------------------------
in your .login script (if you use the csh). An equivalent statement
will adjust the ctype locale for non-csh users.

The process is the same for other operating systems, e.g. on HP/UX use
'setenv LANG german.iso88591'; on IRIX 5.2 use 'setenv LANG de'; on Ultrix 4.3
use 'setenv LANG GER_DE.8859' and on OSF/1 use 'setenv LANG
de_DE.88591'.  The examples given here are for German.  Other
languages work too, depending on your operating system.  Check out
'man setlocale' on your system for more information.

Footnote on HP/UX systems:
As of 10.0, you can use either german.iso88591 or de_DE.iso88591 (a
name more in line with other vendors and developing standards for
locale names).  For a complete listing of locale names, see the text
file /usr/lib/nls/config.  Or, on HP-UX 10.0, execute locale -a . This
command will list all locales currently installed on your system.

4. Selecting the right font under X-11 for xterm (and other applications)
To actually display accented characters, you need to select a font
which does contains bit maps for ISO 8859-1 characters in the
correct character positions. The names of these fonts normally
have the suffix "iso8859-1". Use the command
# xlsfonts
to list the fonts available on your system. You can preview a
particular font with the
# xfd -fn <fontname>
command.

Add the appropriate font selection to your ~/.Xdefaults file, e.g.:
----------------------------------------------------------------------------
XTerm*Font: -adobe-courier-medium-r-normal--18-180-75-75-m-110-iso8859-1
Mosaic*XmLabel*fontList: -*-helvetica-bold-r-normal-*-14-*-*-*-*-*-iso8859-1
----------------------------------------------------------------------------

Footnote: The X11R5 distribution has some fonts which are labeled as
ISO fonts, but which contain only the US-ASCII characters.

5. Translating between different international character sets.
While ISO 8859-1 is an international standard, not everybody uses this
encoding. Many computers use their own, vendor-specific character sets
(most notably Microsoft for MS-DOS).  If you want to edit or view files
written in different encoding, you will have to translate them to an
ISO 8859-1 based representation.

There are several PD/free character set translators available on the
Internet, the most notable being 'recode'.  recode is available from
URL ftp://prep.ai.mit.edu/u2/emacs.  recode is covered by FSF
copyright and is freely redistributable.  

The general format of the program call is one of:

recode [OPTION]... [BEFORE]:[AFTER] [FILE]

The second form is the common case.  Each FILE will be read assuming
it is coded with charset BEFORE, it will be recoded over itself so to
use the charset AFTER.  If there is no such FILE, the program rather
acts as a filter and recode standard input to standard output.

Some recodings are not reversible, so after you have converted the
file (recode overwrites the original file with the new version!), you
may never be able to recontruct the original file.  A safer way of
changing the encoing of a file is to use the filter mechanism of
recode and invoke it as follows:

recode [OPTION]... [BEFORE]:[AFTER] <[OLDFILE] >[NEWFILE]

Under SunOS, the dos2unix and unix2dos programs (distributed with
SunOS) will translate between MS-DOS and ISO 8859-1 formats.

It is somewhat more difficult to convert German, `Duden'-conformant
Ersatzdarstellung (? = ae, ? = sz (or not so conformant `ss') etc.)
into the ISO 8859-1 character set.  The German dictionary available as
URL ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/dicts/deutsch.tar.gz also
contains a UNIX shell script which can handle all conversions except
ones involving ? (German scharfes-s), as for `ss' this change is more
complicated.

A more sophisticated program to translate Duden Ersatzdarstellung to
ISO 8859-1 is Gustaf Neumann's diac program (version 1.3 or later)
which can translate all ASCII sequences to their respective ISO 8859-1
character set representation.  'diac' is available in URL
ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/diac.

Translating ISO 8859-1 to ASCII can be performed with a little sed
script according to your needs.  But be aware that
* No one-to-one mapping between Latin 1 and ASCII strings is possible.
* Text layout may be destroyed by multi-character substitutions,
  especially in tables.
* Different replacements may be in use for different languages,
  so no single standard replacement table will make everyone happy.
* Truncation or line wrapping might be necessary to fit textual data
  into fields of fixed width.
* Reversing this translation may be difficult or impossible.
* You may be introducing ambiguities into your data.

6. Printing accented characters.

6.1 PostScript printers
If you want to print accented characters on a postscript printer, you
may need a PS filter which can handle ISO characters.

Our Postscript filter of choice is a2ps, the more recent version of
which can handle ISO 8859-1 characters with the -8 option.  a2ps V4.3
is available as URL ftp://imag.imag.fr/archive/postscript/a2ps.V4.3.tar.gz.

If you use the pps postscript filter, use the 'pps -ISO' option for
pps to handle ISO 8859-1 characters properly.

6.2 Other (non-PS) printers:
If you want to print to non-PS printers, your success rate depends on
the encoding the printer uses. Several alternatives are possible:

* Your printer accepts ISO 8859-1:
  You're lucky. No conversion is needed, just send your files to the
  printer.

* You printer supports a PC-compatible font:
  You can use the recode tool to translate from ISO 8859-1 to this
  encoding. (If you are using a SunOS based computer, you can also use
  the unix2dos utility which is part of the standard distribution.)
  Just add the appropriate invocation as a built-in filter to your
  printer driver.  

* Your printer uses a national ISO 646 variant (7 bit ASCII
  with some special characters replaced by national characters):
  You will have to use a translation tool; this tool would
  then be installed in the printer driver and translate character
  conventions before sending a file to the printer.  The recode
  program supports many national ISO 646 norms.  (If you add do
  this, please submit it to the maintainers of recode, so that it can
  benefit everybody.)

  Unfortunately, you will not be able to display all characters with
  the built-in characters set. Most printers have user-definable
  bit-map characters, which you can use to print all ISO characters.
  You just have to generate a pix-map for any particular character and
  send this bitmap to the printer.  The syntax for these characters
  varies, but a few conventions have gained universal acceptance
  (e.g., many printers can process Epson-compatible escape sequences).

* Your printer supports a strange format:
  If your printer supports some other strange format (e.g. HP Roman8,
  DEC MCS, Atari, NeXTStep, EBCDIC or what have you), you have to add a
  filter which will translate ISO 8859-1 to this encoding before
  sending your data to the printer.  'recode' supports many of these
  character sets already.  If you have to write your own conversion
  tool, consider this as a good starting base. (If you add support for
  any new character sets, please submit your code changes to the
  maintainers of recode).

  If your printer supports DEC MCS, this is nearly equivalent to ISO
  8859-1 (actually, it is a former ISO 8859-1 draft standard. The only
  characters which are missing are the Icelandic characters (eth and
  thorn) at locations 0xD0, 0xF0, 0xDE and 0xFE) - the difference is
  only a few characters.  You could probably get by with just sending
  ISO 8859-1 to the printer.

* Your printer supports ASCII only:
  You have several options:
  + If your printer supports user-defined characters, you can print all
    ISO characters not supported by ASCII by sending the appropriate
    bitmaps.  You will need a filter to convert ISO 8859-1 characters
    to the appropriate bitmaps.  (A good starting point would be recode.)
  + Add a filter to the printer driver which will strip the accent
    characters and just print the unaccented characters. (This
    character set is supported by recode under the name `flat' ASCII.)
  + Add a filter which will generate escape sequences (such as
    " <BACKSPACE> a for Umlaut-a (?), etc.) to be printed.  Recode
    supports this encoding under the name `ascii-bs'.

Footnote: For more information on character translation and the
'recode' tool, see section 5.

7. TeX and ISO 8859-1
If you want to write TeX without having to type {\"a}-style escape
sequences, you can either get a TeX versions configured to read 8-bit
ISO characters, or you can translate between ISO and TeX codings.

The latter is arduous if done by hand, but can be automated if you use
emacs. If you use Emacs 19.23 or higher, simply add the following line
to your .emacs startup file. This mode will perform the necessary
translations for you automatically:
------------------
(require 'iso-cvt)
------------------

If you are using pre-19.23 versions of emacs, get the "gm-lingo.el"
lisp file via URL ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit.  Load
gm-lingo from your .emacs startup file and this mode will perform the
necessary translations for you automatically.

If you want to configure TeX to read 8 bit characters, check out the
configuration files available in URL
ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit.

In LaTeX 2.09 (or earlier), use the isolatin or isolatin1 styles to
include support for ISO latin1 characters.  Use the following
documentstyle definition:
\documentstyle[isolatin]{article}

isolatin.sty and isolatin1 are available from all CTAN servers and
from URL ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit. (The isolatin1
version on vlsivie is more complete than the one on CTAN servers.)

There are several possibilities LaTeX 2e to provide comprehensive
support for 8 bit characters:

The preferred method is to use the inputenc package with the latin1
option.  Use the following package invocation to achieve this:
\usepackage[latin1]{inputenc}

Alternatively, the styles used for earlier LaTeX versions (see above)
can also be used with 2e.  To do this, use the commands:
\documentclass{article}
\usepackage{isolatin}

You can also get the latex-mode to handle opening and closing quotes
correctly for your language.  This can be achieved by defining the
emacs variables 'tex-open-quote' and 'tex-closing-quote'.  You can
either set these varaibles in your ~/.emacs startup file or as a
buffer-local variable in your TeX file if you want to define quotes on
a per-file basis.

For German TeX quotes, use:
-----------
(setq tex-open-quote "\"`")
(setq tex-closing-quote "'\"")
-----------

If you want to use French quotes (guillemets), use:
-----------
(setq tex-open-quote "?")
(setq tex-closing-quote "?")
-----------

8. ISO 8859-1 and emacs
Emacs 19 (as opposed to Emacs 18) can automatically handle 8 bit
characters. (If you have a choice, upgrade to Emacs version 19.23,
which has the most complete ISO support.)  Emacs 19 has extensive
support for ISO 8859-1. If your display supports ISO 8859-1 encoded
characters, add the following line to your .emacs startup file:
-----------------------------
(standard-display-european t)
-----------------------------

If want to display ISO-8859-1 encoded files by using TeX-like escape
sequences (e.g. if your terminal supports only ASCII characters), you
should  add the following line to your .emacs file (DON'T DO THIS IF
YOUR TERMINAL SUPPORTS ISO OR SOME OTHER ENCODING OF NATIONAL
CHARACTERS):
--------------------
(require 'iso-ascii)
--------------------

If your terminal supports a non-ISO 8859-1 encoding of national
characters (e.g. 7 bit national variant ISO 646 character sets,
aka. `national ASCII' variants), you should configure your own display
table.  The standard emacs distribution contains a configuration
(iso-swed.el) for terminals which have ASCII in the G0 set and a
Swedish/Finnish version of ISO 646 in the G1 set.  If you want to
create your own display table configuration, take a look at this
sample configuration and at disp-table.el for available support
functions.

Emacs can also accept 8 bit ISO 8859-1 characters as input. These
character codes might either come from a national keyboard (and
driver) which generates ISO-compliant codes, or may have been entered
by use  of a COMPOSE-character mechanism.
If you use such an input format, execute the following expression in
your .emacs startup file to enable Emacs to understand them:
-------------------------------------------------
     (set-input-mode (car (current-input-mode))
                     (nth 1 (current-input-mode))
                     0)
-------------------------------------------------

In order to configure emacs to handle commands operating on words
properly (such as 'Beginning of word, etc.), you should also add the
following line to your .emacs startup file:
-------------------------------
(require 'iso-syntax)
-------------------------------

For further information on using ISO 8859-1 with emacs, also see the
Emacs manual section on "European Display" (available as hypertext
document by typing C-h i in emacs or as a printed version).

9. Typing ISO with US-style keyboards.  
Many computer users use US-ASCII keyboards, which do not have keys for
national characters.  You can use escape sequences to enter these
characters.  For ASCII terminals (or PCs), check the documentation of
your terminal for particulars.

9.1 US-keyboards under X11
Under X Windows, the COMPOSE multi-language support key can be used to
enter accented characters.  Thus, when running X11 on a SunOS-based
computer (or any other X11R4 or X11R5 server supporting COMPOSE
characters), you can type three character sequences such as
COMPOSE " a -> ?
COMPOSE s s -> ?
COMPOSE ` e ->
to type accented characters.

Note that this COMPOSE capability has been removed as of X11R6,
because it does not adequately support all the languages in the world.
Instead, compose processing is supposed to be performed in the client
using an `input method', a mechanism which has been available since
X11R5.  (In the short term, this is a step backward for European
users, as few clients support this type of processing at the moment.
It is unfortunate that the X Consortium did not implement a mechanism
which allows for a smoother transition.  Even the xterm terminal
emulator supplied by the X Consortium itself does not yet support this
mechanism!)

Input methods are controlled by the locale environment variables (LANG
and LC_xxx).  The values for these variables are (or at least, should be
made equivalent by any sane vendor) equivalent to those expected by
the ANSI/POSIX locale library.  For a list of possible settings see
section 3.

9.2 US-keyboards and emacs There are several modes to enter Umlaut
characters under emacs when using a US-style keyboard.  One such mode
is iso-transl, which is distributed with the standard emacs
distribution.  This mode uses the Alt-key for entering diacritical
marks (accents et al.).  An extended iso-transl mode (iso-transl+)
which allows the definition of language specific short cuts is
available as URL ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/iso-transl+.shar.  
This file also includes sample configurations for the German and
Spanish languages.

An alternative to using Alt-sequences for entering diacritical marks
is the use of `electric accents', such as used on old type writers or
under many MS Windows programs.  With this method, typing an accent
character will place this accent on the next character entered.  One
mode which supports this entry method is the iso-acc minor mode which
comes with the standard emacs distribution. Just add
------------------
(require 'iso-acc)
------------------
to your emacs startup script, and you can turn the '`~/^" keys into
electric accents by typing 'M-x iso-accents-mode' in a specific
buffer.  To type the ? (c with cedilla) and ? (German scharfes s)
characters, type ~c and "s, respectively.

Footnote: When starting up under X11, Emacs looks for a Meta key and
if it finds no Meta key, it will use the Alt key instead.  The way to
solve this problem, is to define a Meta key using the xmodmap utility
which comes with X11.

10. File names with ISO characters
If your OS is 8 bit clean, you can use ISO characters in file names.
(This is possible under SunOS.)

11. Command names with ISO 8859-1
If your OS supports file names with ISO characters, and your shell is
8 bit clean, you can use command names containing ISO characters. If
your shell does not handle ISO characters correctly, use one of the
many PD shells which do (e.g. tcsh, an extended csh). These are
available from a multitude of ftp sites around the world.

See section 14 on application specific information for a discussion of
various shells.

12. Spell checking
Ispell 3.1 has by far the best understanding of non-English
languages and can be configured to handle 8-bit characters
(Thus, it can handle ISO-8859-1 encoded files).

Ispell 3.1 now comes with hash tables for several languages (English,
German, French,...).  It is available via URL ftp://ftp.cs.ucla.edu/pub.  
Ispell also contains a list of international dictionaries and about
their availability in the file ispell/languages/Where.

To choose a dictionary for ispell, use the `-d <dictionary>'
option.  The `-T <input-encoding>' option should be set set to `-T
latin1' if you want to use ISO 8859-1 as input encoding.

If you use ispell inside emacs (using the ispell.el mode) to spell
check a buffer, you can choose language and input encoding either
using the `M-x ispell-change-dictionary' function, or by choosing the
`Spell' item in the `Edit' pull-down menu.  This will present you with
a choice of dictionaries (cum input encodings): all languages are
listed twice, such as in `Deutsch' and `Deutsch8'.  `Deutsch8' is the
setting which will use the German dictionary and the 8 bit ISO 8859-1
input encoding.  

Alternatively, ispell.el lets you specify the dictionary to use for a
particular file at the end of of that file by adding a line such as
----
Local IspellDict: castellano8
----

The following sites also have dictionaries for ispell available via
anonymous ftp:
language        site                     file name
French          ireq-robot.hydro.qc.ca   /pub/ispell
French          ftp.inria.fr             /INRIA/Projects/algo/INDEX/iepelle
French          ftp.inria.fr             /gnu/ispell3.0-french.tar.gz
German          ftp.vlsivie.tuwien.ac.at /pub/8bit/dicts/deutsch.tar.gz
Spanish         ftp.eunet.es             /pub/unix/text/TeX/spanish/ispell

Some spell checkers use strange encodings for accented characters.  If
you have to use one of these spell checkers, you may have to run
recode before invoking the spell checker to generate a file using your
spell checker's coding conventions.  After running the spell checker,
you have to translate the file back to ISO with recode.

Of course, this can be automated with a shell script:
---------------------
recode <options to generate spell checker encoding from ISO> $i tmp.file
spell_check tmp.file
recode <options to generate ISO from spell checker encoding> tmp.file $i
---------------------

Footnote: Ispell 4.* is not a superset of ispell 3.*. Ispell 4.* was
          developed independently from a common ancestor, but DOES NOT
          support any internationalization, but is restricted to the
          English language.

13. TCP and ISO 8859-1
TCP was specified by US-Americans, for US-Americans. TCP still carries
this heritage: while TCP/IP protocol itself *is* 8 bit clean, no
effort was made to support the transfer of non-English characters in
many application level protocols (mail, news, etc.).  Some of these
protocols still only specify the transfer of 7-bit data, leaving
anything else implementation dependent.

Since the TCP/IP protocol itself transfers 8 bit data correctly,
writing applications based on TCP/IP does not lead to any loss of
encoding information.

13.1 FTP and ISO 8859-1
FTP has support for transferring 8 bit binary data. This mode should be
used when transferring ISO coded data between two hosts.  This mode is
normally enabled by the command "binary".

Note, however, that use of the binary mode for text files will disable
translation between the line-ending conventions of different operating
systems.  You might have to provide some filter to convert between the
LF-only convention of Unix and the CR-LF convention of VMS and MS
Windows when you copy from one of these systems to another.

13.2 Mail and ISO 8859-1
Most Internet eMail standards come from a time when the Internet was a
mostly-US phenomenon.  Other countries did have access to the net, but
much of the communication was in English nevertheless.  With the
propagation of Internet, these standards have become a problem for
languages which cannot be represented in a 7 bit ISO 646 character
set.  

Using ISO 646, which uses a slightly different character set for each
language, also poses a problem when crossing a language barrier, as
the interpretation of characters will change.  As a result, most
countries use the ISO 646 standard commonly referred to as US-ASCII
and will use escape sequences such as 'e () or "a (?) to refer to
national characters.  The exception to this rule are Nordic countries
(more so in Sweden and Finland, less so in Denmark and Norway, I'm
being told), where the national ISO 646 variant has garnered a
formidable following and is a common reference point for all Nordic
users.

As this situation is clearly unsatisfactory, several methods of
sending mails encoded in national character sets have been developed.
We start with a discussion of the mail delivery infrastructure and
will then look at some high-level protocols which can protect mail
users and their messages from the shortcomings of the underlying mail
protocols.

Footnote: Many other email standards exist for proprietary systems.
If you use one of these mail systems, it is the responsibility of the
mail gateway to translate your messages to an appropriate Internet
mail message when you send a message to the Internet.

13.2.1 Mail Transfer Agents and the Internet Mail Infrastructure
The original sendmail protocol specification (SMTP) in RFC 821
specified the transfer of only 7 bit messages.  Many sendmail
implementations have been made 8 bit transparent (see RFC 1428), but
some SMTP handling agents are still strictly conforming to the
(somewhat outdated) RFC 821 and intentionally cut off the 8th bit.
This behavior stymies all efforts to transfer messages containing
national characters.  Thus, only if all SMTP agents between mail
originator and mail recipient are 8 bit clean, will messages be
transferred correctly.  Otherwise, accented characters are mapped to
some ASCII character (e.g. Umlaut a -> 'd'), but the rest of the
messages is still transferred correctly.

A new, enhanced (and compatible) SMTP standard, ESMTP, has been
released as RFC 1425. This standard defines and standardizes 8 bit
extensions.  This should be the mail protocol of choice for newly
shipped versions of sendmail.

Much of the European and Latin American network infrastructure
supports the transfer of 8 bit mail messages, the success rate is
somewhat lower for the US.

DEC Ultrix sendmail still implements the somewhat outdated RFC 821 to
the letter, and thus cuts off the eighth bit of all mail passing
through it.  Thus ISO encoded mail will always lose the accent marks
when transferred through a DEC host.

If your computer is running DEC Ultrix and you want it to handle 8 bit
characters properly, you can get the source for a more recent version
of sendmail via ftp (see section 14.9).  OR, you can simply
call DEC, complain that their standard mail system cannot handle
international 8 bit mail, encourage them to implement 8 bit
transparent SMTP, or (even better) ESMTP, and ask for the sendmail
patch which makes their current sendmail 8 bit transparent.
(Reportedly, such a patch is available from DEC for those who ask.)
In the meantime, an 8 bit transparent sendmail MIPS binary for Ultrix
is available as URL
ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/mips.sendmail.8bit)

If you want to change MTAs, the popular smail PD-MTA is also 8 bit
clean.

13.2.2 High-level protocols
In the Good Old Days, messages were 7-bit US-ASCII only. When users
wanted to transfer 8 bit data (binaries or compressed files, for
example), it was their responsibility to translate them to a 7 bit
form which could be sent.  At the other end, the recipient had to
unpack the data using the same protocol.  The commonly used encoding
mechanism used for this purpose is uuencode/uudecode.

Today, a standard, MIME (MIME stands for Multi-purpose Internet Mail
Extensions), exists which automatically packs and unpacks data as is
required.  This standard can take advantage of different underlying
protocol capabilities and automatically transform messages to
guarantee delivery.  This standard can also be used to include
multimedia data types in your mail messages.

The MIME standard defines a mail transfer protocol which can handle
different character sets and multimedia mail, independent of the
network infrastructure.  This protocol should eventually solve
problems with 7-bit mailers etc.  Unfortunately, no mail transfer
agents (mail routers) and few end user mail readers support this
standard.  Source for supporting MIME (the `metamail' package) in
various mail readers is available in URL
ftp://thumper.bellcore.com/pub/nsb.  MIME is specified in RFC 1521 and
RFC 1522 which are available from ftp.uu.net.  There is also a MIME
FAQ which is available as URL
ftp://ftp.ics.uci.edu/mh/contrib/multimedia/mime-faq.txt.gz.  (This
file is in compressed format. You will need the GNU gunzip program to
decompress this file.)

PS: Newer versions of sendmail support ESMTP negotiation and can pass
8 bit data.  However, they do not (yet?)  support downgrading of 8 bit
MIME messages.

13.3 News and ISO 8859-1
Much as mail, the Usenet news protocol specification is 7 bit based,
but a significant part of the infrastructure has recently been
upgraded to 8 bit service...  Thus, accented characters are transferred
correctly between much of Europe (and Latin America), but accents
sometimes get lost in networks which run old news software (BNews).

ISO 8859-1 is _the_ standard for typing accented characters in most
newsgroups (may be different for MS-DOS centered newsgroups ;-), and
is preferred in most European news group hierarchies, such as at.* or
de.*

For those who speak French, there is an excellent FAQ on using ISO
8859-1 coded characters on Usenet by Fran?ois Yergeau.  This FAQ is
regularly posted in soc.culture.french and other relevant newsgroups.

13.4 WWW (and other information servers)
The WWW protocol can transfer 8 bit data without any problems and you
can advertise ISO-8859-1 encoded data from your client.  The display
of data is dependent upon the user client.  xmosaic (freely available
from the NCSA) which is available for most UNIX platforms uses an
ISO-8859-1 compliant font by default and will display data correctly.

13.5 rlogin
For rlogin to pass 8 bit data correctly, invoke it with 'rlogin -8' or
'rlogin -L'.

14. Some applications and ISO 8859-1
14.1 bash
You need version 1.13 or higher and set the locale correctly (see
section 3).  Also, to configure the `readline' input function of bash
to handle 8 bit characters correctly, you have to set some environment
variables in the readline startup file .inputrc:
-------------------------------------------------------
set meta-flag On
set convert-meta Off
set output-meta On
-------------------------------------------------------

Before bash version 1.13, bash used the eighth bit of characters to
mark whether or not they were quoted when performing word expansions.
While this was not a problem in a 7-bit US-ASCII environment, this was
a major restriction for users working in a non-English environment.

These readline variables have the following meaning (and default
values):
     meta-flag (Off)
          If set to On,  readline  will  enable  eight-bit  input
          (that is, it will not strip the high bit from the char-
          acters it  reads),  regardless  of  what  the  terminal
          claims it can support.
     convert-meta (On)
          If set to On, readline will convert characters with the
          eighth  bit  set  to an ASCII key sequence by stripping
          the eighth bit and prepending an escape  character  (in
          effect, using escape as the meta prefix).
     output-meta (Off)
          If set to On, readline will display characters with the
          eighth  bit set directly rather than as a meta-prefixed
          escape sequence.

Bash is available from prep.ai.mit.edu in /pub/gnu.

14.2 elm
Elm automatically supports the handling of national character sets,
provided the environment is configured correctly.  If you configure
elm without MIME support, you can receive, display, enter and send 8
bit ISO 8859-1 messages (if your environment supports this character
set).

When you compile elm with MIME support, you have two options:
* you can compile elm to use 8 bit ISO-8859-1 as transport encoding:
  If you use this encoding even people without MIME compliant mailers
  will be able to read your mail messages, if they use the same
  character set.  The eight bit may however be cut off by 7 bit MTAs
  (mail transfer agents), and mutilated mail might be received by the
  recipient, regardless of whether she uses MIME or not.  (This
  problem should be eased when 8 bit mailers are upgraded to
  understand how to translate 8 bit mails to 7 bit encodings when they
  encounter a 7 bit mailer.)

* you can compile elm to use 7 bit US-ASCII `quoted printable' as
  transport encoding:
  this encoding ensures that you can transfer your mail containing
  national characters without having to worry about 7 bit MTAs.  A
  MIME compliant mail reader at the other end will translate your
  message back to your national character set.  Recipients without
  MIME compliant mail readers will however see mutilated messages:
  national characters will have been replaced by sequences of the type
  '=FF' (with FF being the ISO code (in hexadecimal) of the national
  character being encoded).

14.3 GNUS
GNUS is a newsreader based on emacs.  It is 8 bit transparent and
contains all national character support available in emacs 19.

14.4 less
Set the LESSCHARSET environment variable with
'setenv LESSCHARSET latin1'.

14.5 metamail
To configure the metamail package for ISO 8859-1 input/output, set the
MM_CHARSET environment variable with 'setenv MM_CHARSET ISO-8859-1'.
Also, set the MM_AUXCHARSETS variable with 'setenv MM_AUXCHARSETS
iso-8859-1'.

14.6 nn
Add the line
-----------------
set data-bits 8
-----------------
to your ~/.nn/init (or the global configuration file) in order for nn
to be able to process 8 bit characters.

14.7 nroff
The GNU replacement for nroff, groff, has an option to generate ISO
8859-1 coded output, instead of plain ASCII.  Thus, you can preview
nroff documents with correctly displayed accented characters.  Invoke
groff with the 'groff -Tlatin1' option to achieve this.

Groff is free software. It is available from URL
ftp://prep.ai.mit.edu/pub/gnu and many other GNU archives around the
world.

14.8 pgp
PGP (Phil Zimmermann's Pretty Good Privacy) uses Latin1 as canonical
form to transmit crypted data.  Your host computer's local character
set should be configured in the configuration file
${PGPPATH}/config.txt by setting the CHARSET parameter.  If you are
using ISO 8859-1 as your native character set, CHARSET should bet set
to LATIN1, on MS-DOS computers with code page 850 set 'CHARSET =
CP850'.  This will make PGP automatically translate all crypted texts
from/to the LATIN1 canonical form.  A setting of 'CHARSET = NOCONV'
can be used to inhibit all translations.  (

When PGP is used to code Cyrillic text, KOI8 is regarded as canonical
form (use 'CHARSET = KOI8').  If you use the ALT_CODES encoding for
Cyrillic (popular on PCs), set 'CHARSET = ALT_CODES' and it will
automatically be converted to KOI8.

Footnote: Note that PGP treats KOI8 as LATIN1, even though it is a
completely different character set (Russian), because trying to
convert KOI8 to either LATIN1 or CP850 would be futile anyway.

14.9 sendmail
BSD Sendmail Version 8 has a flag in the configuration file set to
True or False which determines whether v8 passes any 8-bit data it
encounters, presumably to match the behavior of other 8-bit
transparent MTAs and to meet the wants of non-ASCII users, or if it
strips to 7 bits to conform to SMTP.  The source code for an 8 bit
clean sendmail is available in URL ftp://ftp.cs.berkeley.edu/ucb/sendmail.  
A pre-compiled binary for DEC MIPS systems running Ultrix is available
as URL ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/mips.sendmail.8bit.

14.10 tcsh
You need version 6.04 or higher, and your locale has to be set
properly (see section 3).  Tcsh also needs to be compiled with the
national language support feature, see the config.h file in the tcsh
source directory.  Tcsh is an extended csh and is available in URL
ftp://tesla.ee.cornell.edu/pub/tcsh.

14.11 vi
Support for 8 bit character sets depends on the OS.  It works under
SunOS 4.1.*, but on OSF/1 vi gets confused about the current cursor
position in the presence of 8 bit characters.

15. Terminals
15.1 X11 Terminal Emulators
15.1.1 xterm
If you are using X11 and xterm as your terminal emulator, you should
place the following line in ~/.Xdefaults (this seems to be required in
some releases of X11, not in all):
-------------------------
XTerm*EightBitInput: True
-------------------------

15.1.2 rxvt
rxvt is another terminal emulator used for X11, mostly under
Linux.  Invoke rxvt with the 'rxvt -8' command line.

15.2 VT2xx, VT3xx
The character encoding used in VT2xx terminals is a preliminary
version of the ISO-8859-1 standard (DEC MCS), so some characters (the
more obscure ones) differ slightly.  However, these terminals can be
used with ISO 8859-1 characters without problems.

The newer VT3xx terminals use the official ISO 8859-1 standard.

The international versions of the VT[23]xx terminals have a COMPOSE
key which can be used to enter accented characters, e.g.
<COMPOSE><e><'> will give an e with accent aigu ().

15.3 Various UNIX terminals
Some terminals support down-loadable fonts.  If characters sent to
these terminals can be 8 bit wide, you can down-load your own ISO
characters set.  To see how this can be achieved, take a look at the
/pub/culture/russian/comp/cyril-term on nic.funet.fi.

15.4 MS-DOS PCs
MS-DOS PCs normally use a different encoding for accented characters,
so there are two options:

* you can use a terminal emulator which will translate between the
  different encodings.  If you use the PROCOMM PLUS, TELEMATE and
  TELIX modem programs, you can down-load the translation tables
  from URL ftp://oak.oakland.edu/pub/msdos/modem/xlate.zip.

* you can reconfigure your MS-DOS PC to use an ISO-8859-1 code page.
  Either install IBM code page 819 (see section 19), or you can get
  the free ISO 8859-X support files from the anonymous ftp archive
  ftp.uni-erlangen.de, which contains data on how to do this (and
  other ISO-related stuff) in /pub/doc/ISO/charsets.  The README file
  contains an index of the files you need.

16. Programming applications which support the use of ISO 8859-1
For information on how to write applications with support for
localization (to the ISO 8859-1 and other character representations)
check out URL ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/ISO-programming.

17. Other relevant i18n FAQs
This is a list of other FAQs on the net which might be of interest.
Topic                   Newsgroup(s)            Comments
Nordic graphemes        soc.culture.nordic      interesting stuff about
                                                handling nordic letters
accents sur Usenet      soc.culture.french,...  Accents on Usenet (French)
                                                + more
Programming for I18N    comp.unix.questions,... see section 16.
International fonts     ftp://ftp.vlsivie.tuwien.ac.at/pub/8bit/ISO-fonts
                                                Discusses international fonts
                                                and where to find them

18. Operating Systems and ISO 8859-1
18.1 UNIX
Most Unix implementations use the ISO 8859--1 character set, or at
least have an option to use it.  Some systems may also support other
encodings, e.g.~Roman8 (HP/UX), DEC MCS (DEC Ultrix, see the section
on VMS), etc.

18.2 NeXTSTEP
NeXTSTEP uses a proprietary character set.

18.3 MS DOS
IBM code page 819 _is_ ISO 8859-1.  Code Page 850 has the same
characters as ISO 8859-1, BUT the characters are in different
locations (i.e., you can translate 1-to-1, but you do have to
translate the characters.)

18.4 MS-Windows
Microsoft Windows uses an ISO 8859-1 compatible character set (Code
Page 1252), as delivered in the US, Europe (except Eastern Europe) and
Latin America.  In Windows 3.1, Microsoft has added additional characters
in the 0x80-0x9F range.

18.5 DEC VMS
DEC VMS uses the DEC MCS character set, which is practically
equivalent to ISO 8859-1 (it is a fromer ISO 8859--1 draft standard).
The only characters which differ between DEC MCS and ISO 8859-1 are
the Icelandic characters (eth and thorn) at locations 0xD0, 0xF0, 0xDE
and 0xFE.

19. Table of ISO 8859-1 Characters
This section gives an overview of the ISO 8859-1 character set.  The
ISO 8859-1 character set consists of the following four blocks:

00      19      CONTROL CHARACTERS
20      7E      BASIC LATIN
80      9F      EXTENDED CONTROL CHARACTERS
A0      FF      LATIN-1 SUPPLEMENT

The control characters and basic latin blocks are similar do those
used in the US national variant of ISO 646 (US-ASCII), so they are not
listed here.  Nor is the second block of control characters listed,
for which not functions have yet been defined.  

+----+-----+---+------------------------------------------------------
|Hex | Dec |Car| Description ISO/IEC 10646-1:1993(E)
+----+-----+---+------------------------------------------------------
|    |     |   |
| A0 | 160 |   | NO-BREAK SPACE
| A1 | 161 | ? | INVERTED EXCLAMATION MARK
| A2 | 162 | | CENT SIGN
| A3 | 163 | | POUND SIGN
| A4 | 164 | | CURRENCY SIGN
| A5 | 165 | | YEN SIGN
| A6 | 166 | | | BROKEN BAR
| A7 | 167 | | SECTION SIGN
| A8 | 168 | | DIAERESIS
| A9 | 169 | ? | COPYRIGHT SIGN
| AA | 170 | a | FEMININE ORDINAL INDICATOR
| AB | 171 | ? | LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
| AC | 172 | ? | NOT SIGN
| AD | 173 | - | SOFT HYPHEN
| AE | 174 | ? | REGISTERED SIGN
| AF | 175 | | MACRON
|    |     |   |
| B0 | 176 | | DEGREE SIGN
| B1 | 177 | | PLUS-MINUS SIGN
| B2 | 178 | 2 | SUPERSCRIPT TWO
| B3 | 179 | 3 | SUPERSCRIPT THREE
| B4 | 180 | | ACUTE ACCENT
| B5 | 181 | | MICRO SIGN
| B6 | 182 | ? | PILCROW SIGN
| B7 | 183 | | MIDDLE DOT
| B8 | 184 | ? | CEDILLA
| B9 | 185 | 1 | SUPERSCRIPT ONE
| BA | 186 | o | MASCULINE ORDINAL INDICATOR
| BB | 187 | ? | RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
| BC | 188 | ? | VULGAR FRACTION ONE QUARTER
| BD | 189 | ? | VULGAR FRACTION ONE HALF
| BE | 190 | ? | VULGAR FRACTION THREE QUARTERS
| BF | 191 | ? | INVERTED QUESTION MARK
|    |     |   |
| C0 | 192 | | LATIN CAPITAL LETTER A WITH GRAVE ACCENT
| C1 | 193 | | LATIN CAPITAL LETTER A WITH ACUTE ACCENT
| C2 | 194 | ? | LATIN CAPITAL LETTER A WITH CIRCUMFLEX ACCENT
| C3 | 195 | ? | LATIN CAPITAL LETTER A WITH TILDE
| C4 | 196 | ? | LATIN CAPITAL LETTER A WITH DIAERESIS
| C5 | 197 | ? | LATIN CAPITAL LETTER A WITH RING ABOVE
| C6 | 198 | ? | LATIN CAPITAL LIGATURE AE
| C7 | 199 | ? | LATIN CAPITAL LETTER C WITH CEDILLA
| C8 | 200 | | LATIN CAPITAL LETTER E WITH GRAVE ACCENT
| C9 | 201 | | LATIN CAPITAL LETTER E WITH ACUTE ACCENT
| CA | 202 | | LATIN CAPITAL LETTER E WITH CIRCUMFLEX ACCENT
| CB | 203 | ? | LATIN CAPITAL LETTER E WITH DIAERESIS
| CC | 204 | | LATIN CAPITAL LETTER I WITH GRAVE ACCENT
| CD | 205 | | LATIN CAPITAL LETTER I WITH ACUTE ACCENT
| CE | 206 | ? | LATIN CAPITAL LETTER I WITH CIRCUMFLEX ACCENT
| CF | 207 | ? | LATIN CAPITAL LETTER I WITH DIAERESIS
|    |     |   |
| D0 | 208 | D | LATIN CAPITAL LETTER ETH
| D1 | 209 | ? | LATIN CAPITAL LETTER N WITH TILDE
| D2 | 210 | | LATIN CAPITAL LETTER O WITH GRAVE ACCENT
| D3 | 211 | | LATIN CAPITAL LETTER O WITH ACUTE ACCENT
| D4 | 212 | ? | LATIN CAPITAL LETTER O WITH CIRCUMFLEX ACCENT
| D5 | 213 | ? | LATIN CAPITAL LETTER O WITH TILDE
| D6 | 214 | ? | LATIN CAPITAL LETTER O WITH DIAERESIS
| D7 | 215 | | MULTIPLICATION SIGN
| D8 | 216 | ? | LATIN CAPITAL LETTER O WITH STROKE
| D9 | 217 | | LATIN CAPITAL LETTER U WITH GRAVE ACCENT
| DA | 218 | | LATIN CAPITAL LETTER U WITH ACUTE ACCENT
| DB | 219 | ? | LATIN CAPITAL LETTER U WITH CIRCUMFLEX ACCENT
| DC | 220 | | LATIN CAPITAL LETTER U WITH DIAERESIS
| DD | 221 | Y | LATIN CAPITAL LETTER Y WITH ACUTE ACCENT
| DE | 222 | T | LATIN CAPITAL LETTER THORN
| DF | 223 | ? | LATIN SMALL LETTER SHARP S
|    |     |   |
| E0 | 224 | | LATIN SMALL LETTER A WITH GRAVE ACCENT
| E1 | 225 | | LATIN SMALL LETTER A WITH ACUTE ACCENT
| E2 | 226 | a | LATIN SMALL LETTER A WITH CIRCUMFLEX ACCENT
| E3 | 227 | ? | LATIN SMALL LETTER A WITH TILDE
| E4 | 228 | ? | LATIN SMALL LETTER A WITH DIAERESIS
| E5 | 229 | ? | LATIN SMALL LETTER A WITH RING ABOVE
| E6 | 230 | ? | LATIN SMALL LIGATURE AE
| E7 | 231 | ? | LATIN SMALL LETTER C WITH CEDILLA
| E8 | 232 | | LATIN SMALL LETTER E WITH GRAVE ACCENT
| E9 | 233 | | LATIN SMALL LETTER E WITH ACUTE ACCENT
| EA | 234 | | LATIN SMALL LETTER E WITH CIRCUMFLEX ACCENT
| EB | 235 | ? | LATIN SMALL LETTER E WITH DIAERESIS
| EC | 236 | | LATIN SMALL LETTER I WITH GRAVE ACCENT
| ED | 237 | | LATIN SMALL LETTER I WITH ACUTE ACCENT
| EE | 238 | ? | LATIN SMALL LETTER I WITH CIRCUMFLEX ACCENT
| EF | 239 | ? | LATIN SMALL LETTER I WITH DIAERESIS
|    |     |   |
| F0 | 240 | e | LATIN SMALL LETTER ETH
| F1 | 241 | ? | LATIN SMALL LETTER N WITH TILDE
| F2 | 242 | | LATIN SMALL LETTER O WITH GRAVE ACCENT
| F3 | 243 | | LATIN SMALL LETTER O WITH ACUTE ACCENT
| F4 | 244 | ? | LATIN SMALL LETTER O WITH CIRCUMFLEX ACCENT
| F5 | 245 | ? | LATIN SMALL LETTER O WITH TILDE
| F6 | 246 | ? | LATIN SMALL LETTER O WITH DIAERESIS
| F7 | 247 | | DIVISION SIGN
| F8 | 248 | ? | LATIN SMALL LETTER O WITH OBLIQUE BAR
| F9 | 249 | | LATIN SMALL LETTER U WITH GRAVE ACCENT
| FA | 250 | | LATIN SMALL LETTER U WITH ACUTE ACCENT
| FB | 251 | ? | LATIN SMALL LETTER U WITH CIRCUMFLEX ACCENT
| FC | 252 | | LATIN SMALL LETTER U WITH DIAERESIS
| FD | 253 | y | LATIN SMALL LETTER Y WITH ACUTE ACCENT
| FE | 254 | t | LATIN SMALL LETTER THORN
| FF | 255 | ? | LATIN SMALL LETTER Y WITH DIAERESIS
+----+-----+---+------------------------------------------------------

Footnote: ISO 10646 calls ? a `ligature', but this is supposedly a
          letter in Scandinavian languages.  Thus, it is not in the
          same, merely typographic `ligature' class as `oe' ({\oe} in
          {\LaTeX} convention) which was not included in the ISO
          8859-1 standard.

20. History
In April 1965, the ECMA (European Computer Manufacturer's Association)
stndardized ECMA-6.  This the character set is also (and more
commonly) also know under the names of ISO 646, US-ASCII or DIN 66003.

However, this standard only contained the basic Latin alphabet, with
no provisions for national characters in use all across Europe.  These
characters were later added by replacing several special characters
from the US-ASCII alphabet (such as {[|]}\ etc.).  These variants were
local to each country and were calle `national ISO 646 variants'.
Portability from one country to another was low, as each country had
their own national variant, and some of the special characters were
still needed (such as for programming C), which made this an
altogether unsatisfying solution.

In 1981, IBM released the IBM PC with an 8 bit character set, code
page 437.  The order of the characters added was somewhat confusing,
to say the least.  However, in 1982 the first hardware (DEC VT220 and
VT240 terminal) using a more satisfying character set, the DEC MCS
(Multilanguage Character Set) was released.

This character set was very similar to ISO 6937/2, which is
essentially equivalent to today's ISO 8859-1.  In March 1985, ECMA
standardized ECMA-94, which later came to be known as ISO 8859-1
through 8859-4.  However, ISO 8859-1 was officially stndardized by ISO
only in 1987.

1987 also saw the release of MS-DOS 3.3 which used Code Page 850.
Code Page 850 contains all characters from ISO 8859-1, making a
loss-free conversion possible.  Code Page 819 which was released later
goes one step further, as it is fully ISO 8859-1 compliant.

The ISO 8859-X standard was designed to allow as much interoperability
between character sets as possible.  Thus, all ISO 8859-X character
sets are a superset of US-ASCII and all character sets will render
English text properly.  Also, there is considerable overlap between
several character sets: a text written in German using the ISO 8859-1
character set can be correctly rendered in ISO 8859-2, the Eastern
European character set, where German is the primary foreign language
(-3, -4, -9, -10 supposedly also can display German text without
changes).

While ISO 8859-X was designed for considerable portability, texts are
still restricted mostly to their character set and portability to
other cultural areas is a problem.  One solution is to use a
meta-protocol (such as -> MIME) which specifies the character set
which was used to write a text and which causes the correct character
set to be used in displaying text.

A different approach to overloading the character set as done in the
ISO 8859-X standard (where the locations 0xa0 to 0xff are used to
encode national characters) is to use wider characters.  This is the
approach employed in Unicode (which is an enocing of Basic
MUlitlanguage Plane (BMP) of ISO/IEC 10646).  The downside to this
approach is that most of the software available today only accepts 8
bit wide characters (7 bit if you have bad luck :-( ), so the Unicode
approach is problematic.  This 8 bit restriction permeates nearly all
code in use today, including such system software (file systems,
process identifiers, etc.!).  To ease this problem somewhat, several
representations which map Unicode characters to a variable length 8
bit based encoding have been introduced (this encoding is called
UTF-8).  More information about Unicode can be obtained from URL
http://unicode.org.

21. Glossary: Acronyms, Names, etc.
i18n        I<-- 18 letters -->n = Internationalization
e13n        Europeanization
l10n        Localization
ANSI        American National Standards Institute, the US member of ISO
ASCII       American Standard Code of Information Interchange
CP          Code Page
CP850       Code Page 850, the most widely used MS DOS code page
CR          Carriage Return
CTAN server  ???---a TeX archive server
DEC         Digital Equipment Corp.
DIN         Deutsche Industrie Norm (German Industry Norm)
DOS         Disk Operating System
EBCDIC      ???---a proprietary IBM character set used on mainframes
ECMA        European Computer Manufacturer's Association
emacs       Editing Macros, a family of popular text editors
ESMTP       Enhanced SMTP
Esperanto   A synthetic, ``universal'' language developed by
              Dr.~Zamenhof in~1887.
FSF         Free Software Foundation
FTP         File Transmission Protocol
GNU         GNU's not Unix, an FSF project
HP          Hewlett Packard
HP/UX       HP Unix
IBM         International Business Machines Corp.
IEEE        Institute of Electrical and Electronics Engineers
INRIA       Institut National de Recherche en Informatique et Automation
IP          Internet Protocol
ISO         International Standards Organization
KOI8        ???---a popular encoding for Cyrillic on UNIX workstations
\LaTeX{}    A macro package for \TeX{}
LF          Linefeed
MCS         DEC's Multilingual Character Set---the ISO 8859--1 draft standard
MIME        Multi-Purpose Internet Mail Extension
MS-DOS      Microsoft's program loader
MTA         mail transfer agent
MUA         mail user agent
OS          Operating System
OSF         the Open Software Foundation
OSF/1       the Open Software Foundation's Unix, Revision 1
PGP         Pretty Good Privacy, an encryption package
POSIX       Portable Operating System Interface (an IEEE UNIX standard)
PS          PostScript, Adobe's printer language
RFC         Request for Comment, an Internet standard
sed         stream editor, a UNIX file manipulation utility
SMTP        Simple Mail Transfer Protocol
TCP         Transmission Control Protocol
\TeX{}      Donald Knuth's typesetting program
UDP         User Datagram Protocol
URL         a WWW Uniform Resource Locator
US-ASCII    the US national variant of ISO 646, see ASCII
VMS         ???---DEC's proprietary OS
W3          WWW
WWW         World Wide Web
X11         X Window System

22. Comments
This FAQ is somewhat Sun-centered, though I have tried to include
other machine types.  If you have figured out how to configure your
machine type, please let me (m...@vlsivie.tuwien.ac.at) know so that I
can include it in future revisions of this FAQ.

23. Home location of this document
The most recent version of this document is available via anonymous
ftp from ftp.vlsivie.tuwien.ac.at under the file name
/pub/8bit/FAQ-ISO-8859-1

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

Copyright ? 1994 Michael Gschwind (m...@vlsivie.tuwien.ac.at)

This document may be copied for non-commercial purposes, provided this
copyright notice appears.  Publication in any other form requires the
author's consent.

Dieses Dokument darf unter Angabe dieser urheberrechtlichen
Bestimmungen zum Zwecke der nicht-kommerziellen Nutzung beliebig
vervielf?ltigt werden.  Die Publikation in jeglicher anderer Form
erfordert die Zustimmung des Autors.

Local IspellDict: english

Michael Gschwind, Institut f. Technische Informatik, TU Wien
snail: Treitlstrasse 3-182-2 || A-1040 Wien || Austria
email: m...@vlsivie.tuwien.ac.at   PGP key available via email
phone: +(43)(1)58801 8156          fax: +(43)(1)586 9697