Re: [PATCH v7 0/7] convert: add support for different encodings

Jeff King [off-list ref] writes:

Of the three solutions, I think the relative merits are something like
this:
...
  3. w-t-e (Lars's patch)

I thought Lars's w-t-e was about keeping the in-repo contents in
UTF-8 and externalize in whatever encoding (e.g. UTF-16), so it
won't help the issue hosting folks want to deal with, i.e. showing
in-repo data that is stored in a strange binary-looking encoding in
a more reasonable encodign while diffing, no?

Usually we only work in-repo encoding when producing a diff and show
the result in in-repo encoding, but I can imagine a new attribute,
when set, we first convert in-repo to the specified encoding before
passing the result to xdiff machinery.  Then convert it back to
in-repo encoding before showing the diff (or just show the result in
that encoding xdiff machinery processed---I do not know which one
should be the default).