On Mon, Apr 14, 2008 at 06:42:35PM +0200, Paul van Tilburg wrote:
[...]
It would IMO be more sane to set the default based on the locale.
I have set LC_CTYPE to an UTF-8 locale exactly for that purpose.
If this is done, this still works for editting latin1/latin9 files
or ASCII files.
hmm - it's a little more complicated than that (I've made some changes
which seem to work, will be testing...). For _reading_ a file, the
auto-detection is actually needed, I think, to avoid reading a latin1
file as utf-8 (it would tend to be illegal coding). [...]
For _writing_ a new buffer, the locale is more reliable, since vile's
collecting all of the information on that.
So what I've implemented in the past couple of days is a new setting
for file-encoding, "locale", and made that the default. The existing
"auto" setting tells it to use only its guess, while "locale" tells it
to use the locale-settings for new files. The other values are still
used to tell it exactly what encoding to use.