Re: 8 bit chars on openbsd
Hello Will,
On Tuesday, May 15, 2007 at 16:16:52 -0700, William B. Yardley wrote:
> When sending a message with an accented character in it, mutt (on my
> OpenBSD system) sets the content-type as:
>| Content-Type: text/plain; charset=unknown-8bit
Bad $charset. On OpenBSD 4.0 mutt -v prints +HAVE_LANGINFO_CODESET,
right? Then hopefully a well setup locale should give to Mutt the
appropriate $charset. In the meantime, just hardcode in muttrc:
| set charset=iso-8859-1 # to be removed ASAP
> I tried setting LC_ALL="en_US" and LC_ALL=iso8859-1
First of all: Nobody should never set LC_ALL in normal conditions;
Set LANG instead. And perhaps LC_CTYPE if appropriate on BSDs.
Please try LANG=en_US.ISO8859-1, and check in mutt directly that
":set &charset ?charset" says "iso-8859-1". Otherwise try typing this
also with any other values you might find below. Otherwise try it also
with charset name only: LC_CTYPE=ISO8859-1 The goal is to have Mutt say
"iso-8859-1". Search available values around this:
| $ locale -a
| $ ls /usr/share/locale/ # or such path
Please also report me about this:
| $ locale
| $ printenv | egrep "^(LANG|LC_)" | sort
Westerners can alias "unknown-8bit" to Latin-1 or CP-1252 and have
such broken mails mostly display well:
| charset-hook ^unknown-8bit$ cp1252
| charset-hook ^x-unknown$ cp1252
| charset-hook ^x-user-defined$ cp1252
> Also, when searching for a string, on a linux machine, the regex
> library or something is smart enough to have "e" match an accented e
> as well as a literal e (i.e., a search for "~s stephane" brings up
> subjects with "stéphane" as well).
That sort of things is an advanced feature controlled by the
LC_COLLATE category. The OpenBSD libc has extremely limited or none
locale features. It probably doesn't do that.
BTW: Installing libiconv would give you a way better support of
various charsets in Mutt.
Bye! Alain.
--
Give your computer's unused idle processor cycles to a scientific goal:
The Folding@home project at <URL:http://folding.stanford.edu/>.