> I've posted a patch recently (yesterday) to fix this. The problem is that
> lynx used 'isgraph(c)' for determination of whether the character is not
> blank. Not it will use '!isspace(c)', that should work for any
> charset/encoding. Try that patch if you are curious.
The correct fix for this is to specify the correct locale. Unfortunately
many programs get confused when this is done, so it may be necessary
for Lynx to set just the relevant attributes internally.
(If you set the master locale environment variable, SCO Unix a.out
applications start complaing about broken locale files and some
programs that generate printable numbers can break - e.g. parts of
Hylafax write out decimal numbers for a postscript interpreter and
Postscript uses a locale independent decimal point character (".").)
