Ken Hornstein <kenh@pobox.com> writes:
I take back back what I said, earlier about the default format that my editor
stored nmh bound files in. A few weeks, ago, I did change the default to
UTF-8. But as of, Dec 14 2012, for files whose names are all digits and are in
in my draft folder the default is overridden to ISO-8859-1. I don't know I why
did that.
Well, that would explains things perfectly!
Pending your instructions, I will not fix this problem. The two questions are
what should be the storage charset be for files destined for nmh and should be
the charset for other files.
Nmh assumes that text you give to it is the character set of your locale;
so if you have a UTF-8 locale, it assumes that the text you give it is
UTF-8 and will mark it appropriately. Well, technically if the text is
only ASCII then it will mark it as ASCII, but if there are any 8-bit
characters it will then assume that it's in UTF-8. You can override this
default with an mhbuild directive; if you were to put the following at
the beginning of your text:
#<text/plain; charset=iso-8859-1
and "mime" the draft at the WhatNow prompt, it would label it as ISO-8859-1.
But that would be a pain.
So, what should your editor do? Well, since nmh thinks your character
set is UTF-8 it would make the most sense to have your editor do that
for nmh files.
For your other files ... well, I guess that depends on what you want
to do. My personal feeling is that the world seems to be migrating to
UTF-8 as a default, and unless there is a reason you cannot handle UTF-8
then that's the way you should go. If you've changed your locale to
UTF-8 you've already told the traditional Unix utilities that's what
you're using.
How did you get the saved text from the saved editor file into nmh?
I typed "s\n" into "whatnow". As will do, in a moment, after storing as UTF-8.
I was more thinking about what happened before that; I guess your default
editor that gets invoked by repl is norms-cool-editor?
Correct.
I just changed norms-cool-editor's (yeah! I like that name) default charset,
for writing, to US-ASCII. But files that whose names are all digits and are in
in my draft folder will not be stored as UTF-8.
I am not at all secure about how the standard GNU utilities will handle
non-ascii characters. For example, 'wc -c', just counts bytes. True, the man
page talks about bytes, not characters, but I am still left uncomfortable.
Then there are the dozens of bash, python, and perl scripts that I have
accumulated over the years.
However, if I ever try to store any non-ascii characters, with US-ASCII,
norms-cool-editor will give me the option, of selecting a different charset,
applying a filter that represents non-ascii characters as ascii strings, or
just blindly storing in US-ASCII. I can then make a judgment about what to
do.
Thank you very much.
Norman Shapiro
_______________________________________________
Nmh-workers mailing list
Nmh-workers@nongnu.org
https://lists.nongnu.org/mailman/listinfo/nmh-workers