Re: undo weirdness with insert-file-contents

[martin rudalics <rudalics@gmx.at>]

 > given a 3-way merge conflict, with 3 branches MINE/BASE/OTHER, I do
 > a narrow-to-region on the conflict, then I save all three branches of
 > the conflict to separate files and do:
 >
 >    diff -c -w BASE MINE | patch OTHER >RESULT
 >
 > and if the patch succeeds I do `insert-file-contents' of RESULT without
 > VISIT but with REPLACE.
 >
 > It's indeed somewhat obscure, and indeed, it crashed Emacs when the
 > Unicode branch was merged (because the code didn't expect REPLACE in
 > a narrowed buffer).

With the MINE buffer current, I suppose, to avoid visiting RESULT,
inserting it into MINE, and deleting it afterwards.  As a benefit,
`insert-file-contents' will preserve markers within the conflict region
and store less undo information.  Sounds reasonable.

The examples of non-VISIT&REPLACE I've found in the Emacs sources seem
rather obscure though.