From: Russ Allbery (rra@stanford.edu)
Date: Mon Jan 01 2001 - 13:57:06 CST
Charles Lindsey <chl@clw.cs.man.ac.uk> writes:
> Russ Allbery <rra@stanford.edu> writes:
>> Does the concept of "control character" have a formal definition here?
>> ESC is sometimes used to introduce a character set change; does that make
>> it a control character?
> I think the term is well defined for those character sets that have them.
> If a particular CCS provides for the use of ESC as a means to introduce
> extra glyphs, then I would expect articles written according to that CCS
> to follow its conventions.
> Generally speaking, control characters are defined to "control" things
> other than the transmitted text (such as movements of the carriage on an
> ASR 33, or features of the transmission medium like EOR).
That sounds like an awful lot of handwaving for a standard.
Why don't we just say that if the character set is US-ASCII, the body
should only include a particular set of characters, which we can even give
in ABNF form to be absolutely clear, and that if the character set is
something other than US-ASCII, then the definition of that character set
determines what characters are valid in the body (with the note that any
character set requiring NUL, or CR or LF except as the combined line
ending CRLF must be encoded per MIME)?
Or even better, why don't we bail completely and pass off to MIME, which
already specifies all of this in adequate detail, adding a clarification
about form feed and maybe backspace?
I think we're reinventing wheels already adequately handled by other
standards.
-- Russ Allbery (rra@stanford.edu) <http://www.eyrie.org/~eagle/>