(aao-errors-memo.txta E 941021 OJ)

  Olle Jarnefors                             MEMO
  Royal Institute of Technology, Stockholm                         Version D
  Teknisk Service - Data                     1994-10-21            SUNET-MIME
  08-790 71 26    <ojarnef@admin.kth.se>                           
  Fax:08-10 25 10      Telex:11421 kth s

  
  
! 70 ways (or so) of (mis)representing Swedish letters
  ====================================================
  
  By Olle Jarnefors, TS-Data, KTH (+46-8-790 71 26, <ojarnef@admin.kth.se>)
  
  Summary: There is a great deal of confusion about how to best
  represent the national Swedish letters in Internet email and
  netnews (USENET). Some methods in use and occurring modes of
  misrepresentation are described: 6 character sets, 4 transport
! encodings, 50 transliterations, 10 types of distorted text.

  NOTE: To fully appreciate the content of this file it is
  essential that it is viewed by means of a monospaced font.

: News in this version: See last section.

! Text marked by "!" in the left margin is new or changed
! compared to version D.

: Text marked by ":" in the left margin is new or changed
: compared to version C.

  Content:
  1. Introduction
  2. Direct character representation
  3. Not restored transport encoding
  4. Transliteration to ASCII, often performed manually
  5. Transport-distorted representation
  6. Occasionally observed representations
  7. Overview of the national Swedish letters in common coded character sets
  8. Document history


  1. Introduction
  ---------------
  
  This is an inventoryof the many different ways of
  _representing_ the national Swedish letters that occurs in
  Internet email in Sweden adn articles in netnews (USENET).
  Those letters are:
  
     <A-ring> <A-umlaut> <O-umlaut> <a-ring> <a-umlaut> <o-umlaut>
  
  I have also included the most common forms of _automatic
  distortion_ of these letters one may be exposed to. I have
  myself observed most, if not all, of these in email from
  Internet or articles in netnews.

  To simplify the matter I have disregarded the letters
  <E-acute>, <U-umlaut>, and other letters which are not
  included in the Swedish alphabet but still are sometimes used
  in Swedish text.
  
  It's risky to write about characters that computers and
  networks often fail to represent correctly in a text file made
  available on a public computer network. The _Swedish_ original
  of this text is therefore available in three different
  versions:
  
  aao-fel-pm.txt1   The text is encoded according to the
                    standard ISO 8859-1 (also called Latin-1).
                    This should work without problems in modern
                    Unix systems, MS Windows, MS Windows NT, and
                    OS/2.
  
  aao-fel-pm.txts   The text is encoded by a Swedish 7-bit
                    character set according to the standard
                    SS 63 61 27. This form is recommended by
                    SUNET for use in email.
  
  aao-fel-pm.txta   The text is encoded in (American) ASCII
                    according to the standard ANSI X3.4. THe
                    Swedish national letters are not available
                    and have been replaced by AAO, except where
                    a risk for misunderstanding might occur.
  
  I have allocated unique _o-numbers_ to the different
  representation/distortion modes. The letters are accounted for
  in the order
  
     <A-ring> <A-umlaut> <O-umlaut> <a-ring> <a-umlaut> <o-umlaut>

  Either I write, directly, the sequence of ASCII characters by
  which the letter is represented, or I indicate, indirectly,
  the octet by which is represented, by means of the two
  hexadecimal digits of its value.

  Special marks used:

     Mark      Meaning
     ----      -------
     (+)       common
     <nil>     occurs now and then
     (-)       rare
  
  
  2. Direct character representation
  ----------------------------------
  
  o1 (+)   Swedish 7-bit character sets
              5D  5B  5C  7D  7B  7C
           (hexadecimal representation)
  
  o2 (+)   Latin-1 (= ISO 8859-1), MS Windows character set, DEC MCS
              C5  C4  D6  E5  E4  F6
           (hexadecimal representation)
  
  o3       IBM PC character sets (CP437 and CP850)
              8F  8E  99  86  84  94
           (hexadecimal representation)
  
  o4       Macintosh character sets
              81  80  85  8C  8A  9A
           (hexadecimal representation)
  
  o5 (-)   HP ROMAN-8
              D0  D8  DA  D4  CC  CE
           (hexadecimal representation)
  
  o6 (-)   NeXT, PostScript
              86  85  96  DA  D9  F0
           (hexadecimal representation)
  
  
  3. Not restored transport encoding
  ----------------------------------
  
  o7       Quoted-Printable with Latin-1 as coded character set
              =C5 =C4 =D6 =E5 =E4 =F6
           (teckensekvenser)
  
           Together with, or instead of, these may occur:
              =c5 =c4 =d6 =e5 =e4 =f6
  
  o8 (-)   The so-called mnemonic character sets
              &AA &A: &O: &aa &a: &o:
           (character sequences)
  
           Sometimes the first character of these sequences is instead
  o9       CTRL-] (1D)
  o10      or SP followed by BS (20 08).
  
  
  4. Transliteration to ASCII, often performed manually
  -----------------------------------------------------
  
  o11         A   A   O   a   a   o
  
  Its common to replace the Swedish letters by character pairs
  to decrease the risk of confusion. For <A-ring> and <a-ring>
  often some of the following representations are used:
  
  oa1         AA          aa
  oa2         A*          a*
  oa3         *A          *a
  oa4         A.          a.
  oa5         .a          .a
! oa6         A'          a'
! oa7         'A          'a
  
  For <A-umlaut> <O-umlaut> <a-umlaut> <o-umlaut> often:
  
  oo1             AE  OE      ae  oe
  oo2             A:  O:      a:  o:
  oo3             :A  :O      :a  :o
  oo4             A"  O"      a"  o"
  oo5             "A  "O      "a  "o
  oo6             A%  O%      a%  o%
  oo7             %A  %O      %a  %o
  
  The o-number for combinations of these methods are computed by
  the formulas:
  
!    o_no = 5*(oo_no-1) + oa_no + 11   if        oa_no <= 5
!    o_no = 2*(oo_no-1) + oa_no + 51   if   6 <= oa_no <= 7
  
  
  5. Transport-distorted representation
  -------------------------------------
  
  o47 (+)  EDV-destroyed text (Latin-1 with the 8th bit cleared)
              E   D   V   e   d   v
  
  o48 (-)  Simplest fall-back
              ?   ?   ?   ?   ?   ?
           Other fall-back characters occurs: " ", "_", "#",
           "!", "x", "^"
  
  o49 (-)  All national Swedish letters has disappeared.
  
  o50 (-)  PXZ-distroyed text (the ISO-compatibel coded character
           set ROMAN-8, when the 8th bit has been cleared)
              P   X   Z   T   L   N
  
           When text written with a IBM PC, Macintosh, or NeXT
           character set is mangled by bit-stripping, one gets
           different control characters within the text.
  
  o51 (-)  IBM PC character sets
              0F  0E  19  06  04  14
           (CTRL-O, CTRL-N, CTRL-Y, CTRL-F, CTRL-D, CTRL-T)
  
  o52 (-)  Macintosh character sets
              01  00  05  0C  0A  1A
           (CTRL-A, NUL, CTRL-E, FF, LF, CTRL-Z)
  
  o53 (-)  NeXT, PostScript
              06  05  16  5A  59  70
           (CTRL-F, CTRL-E, CTRL-V, "Z", ";", "p")
  
  
  6. Occasionally observed representations
  ----------------------------------------
  
  o54 (-)  Article <1994May2.171600.5818@lin.foa.se> in swnet.general
                          )   (   !
  
  o55 (-)  Email <199409082211.AAA27130@mail.swip.net>
                          60  5E  5F

: o56 (-)  What was sent from the BBS of the Prime Minister's Office
:          at the beginning of 1994
:                         70  3F  20
  
 
  
  7. Overview of the national Swedish letters in common coded character sets
  --------------------------------------------------------------------------
  
     7  8  P  M  R  N
     =  =  =  =  =  =
  5B [                    7 = Swedish 7-bit character set (SS 63 61 27)
  5C \                    8 = Latin-1 (ISO 8859-1), MS Windows
  5D ]                    P = IBM PC character sets
  7B {                    M = Macintosh character sets
  7C |                    R = ROMAN-8 (HP)
  7D }                    N = NeXT, PostScript
  80          [
  81          ]           ] = <A-ring>
  84       {              [ = <A-umlaut>
  85          \     [     \ = <O-umlaut>
  86       }        ]     } = <a-ring>
  8A          {           { = <a-umlaut>
  8C          }           | = <o-umlaut>
  8E       [              
  8F       ]
  94       |
  96                \
  99       \
  9A          |
  C4    [
  C5    ]
  CC             {
  CE             |
  D0             ]
  D4             }
  D6    \
  D8             [
  D9                {
  DA             \  }
  E4    {
  E5    }
  F0                |
  F6    |


: 8. Document history
: -------------------
:
: C   941014 First English version of the document.
: D   941017 The variant o56 added. Document history added.
! E   941021 New ways of transliterating <A-ring> discovered in
!            soc.culture.nordic, so the number of ways to write
!            Swedish letters has increased to the 70 level.
  
  
  
(aao-errors-memo.txta: END)