I remain concerned that there needs to be more explanation and definition
of the ISO-2022-JP encodings. I would be satisfied with explicit references
to the JIS documents, explicit reference to the ISO-2022 document, and some
definition of what the initial character set and state are for that type.
The following is a definition of the encoding (or codeset) used
in Japanese 821/822 networks.
Announcer sequences ESC 2/0 4/1, ESC 2/0 4/8 and ESC 2/0 4/10
are implicitly assumed. These announcer sequences must not
appear as a part of the data.
(They mean ``Use G0 only and no locking-shifts are allowed,''
``Use 94 characters sets only'' and ``Use 7-bits.'')
Designation sequences ESC 2/8 4/2, ESC 2/8 4/10, ESC 2/4 4/0
and ESC 2/4 4/2 are allowed. Any other designation sequences
are not allowed.
(Each escape sequence designates:
ESC 2/8 4/2: ASCII.
ESC 2/8 4/10: Left half of JIS X0201. (Japanese version of 646)
ESC 2/4 4/0: JIS C6226-1978; so-called ``Old JIS Kanji.''
ESC 2/4 4/2: JIS X0208-1983; so-called ``New JIS Kanji.'')
Any other escape sequences are not allowed.
At the beginning, ESC 2/8 4/2 or ESC 2/8 4/10 may be omitted.
As an exception to 2022, the following rule is applied:
Three bit combinations 0/10, 0/13 and 2/0 can appear only
when a single byte coded character set is designated to G0.
I.e., these bit combinations must not appear when a multi
byte coded character set is designated to G0.
(This rule forces that each line starts and terminates in a
``single byte designation state'' and that the space is handled
as if it is a member of 94 characters single byte graphics sets.)
I don't know this kind of information is appear in a written documents,
but I believe that the above definition can be agreed on between
Japanese experts.