On the other hand, it is arguable that the ``all the world is UNIX''
attitude
is an acceptable one for netnews and there really is no problem with ``just
send 8 bits'' in netnews.
Being a full-time UNIX sysadmin with deep multi-platform roots,
the "just send 8 bits" part of "all the world is UNIX" isn't the problem.
Surely not.
For UNIX, charset labelling causes the real problem.
Have you ever administrated a multi-platform systems with several different
encoding schemes? I have and I know charset labelling is unusable.
What, do you think, happens if a text file is shared by NFS?
I hope you, a UNIX sysadmin, don't say "file type".
The problem is line breaks and other things related to "what's plain text?".
Consider the following suggestions:
Your suggestion is good at most for the simplest processing of ASCII
encoded English text.
* avoid imbedded control characters, including backspace
Do you know we do use escape sequences? If not, see RFC 1468, 1554,
1557.
How do you think about bi-directionality support?
* treat any mixture of any number of TABs and SPACEs
as one instance of "white space"
Have you ever used tbl?
Do you know that some languages, including Japanese, does not
use white spaces at all but make use of tabs.
* consider trailing white space as fair game
for arbitrary appendage or deletion;
it can't be seen, pretend it simply isn't there
It'll make input parser of applications quite complex and is good for
nothing.
* for the sake of those of us with XTs or VT100s,
keep lines shorter than 80 columns, maybe <76
Perhaps, you aren't assuming proportional spacing.
How wide is a column?
Masataka Ohta