How do I disable unwanted character mappings?
 

When I view a man page, or when I vi the output of a man page, or when looking at errors from gcc, I get strange character sequences replacing the ' character.

If from a telnet session I do "man termcap >/tmp/out" and look at an octal dump of /tmp/out I see the sequence

342 200 231

inplace of the ' character, for example (extracts from od -c /tmp/out):

P r o g r a m m e r 342 200 231 s


s e p a r a t e d b y 342 200 231 | 342 200 231

The problem happens when using ssh also, so its not limited to telnet.

How can I disable this conversion behaviour?

I seem to have found the cause.

LANG=en_UK.UTF-8 man termcap
LANG=en_UK.iso8859 man termcap
LANG=en_US.iso8859 man termcap

All display '|' correctly, however

LANG=en_US.UTF-8 man termcap

Displays '|' as \0342\0200\0231|\0342\0200\0231

Why would en_US.UTF-8 locale make it do that?