Mercurial > emacs
changeset 39007:f3d3950ed951
(Finsert_file_contents): Undo change of 2001-08-27.
| author | Gerd Moellmann <gerd@gnu.org> |
|---|---|
| date | Wed, 29 Aug 2001 10:31:51 +0000 |
| parents | 47ac337b9bde |
| children | a2722f5391fe |
| files | src/fileio.c |
| diffstat | 1 files changed, 15 insertions(+), 4 deletions(-) [+] |
line wrap: on
line diff
--- a/src/fileio.c Wed Aug 29 09:31:50 2001 +0000 +++ b/src/fileio.c Wed Aug 29 10:31:51 2001 +0000 @@ -4350,10 +4350,21 @@ || coding.type == coding_type_raw_text)) { /* Visiting a file with these coding systems should always make - the buffer unibyte. The call to Fset_buffer_multibyte - ensures that existing markers etc are converted to unibyte, - too. */ - Fset_buffer_multibyte (Qnil); + the buffer unibyte. If we happen to be replacing text in a + multibyte buffer (this happens when reverting an RMAIL + buffer), positions in the buffer, markers etc. may have byte + positions != character positions, so just setting + enable_multibyte_characters to nil doesn't suffice. */ + if (!NILP (replace) + && !NILP (current_buffer->enable_multibyte_characters)) + Fset_buffer_multibyte (Qnil); + else + /* We cannot Fset_buffer_multibyte(nil) here. When we visit a + file literally in a multibyte buffer, a call to + Fset_buffer_multibyte would change character positions, + although the buffer has unibyte contents, and that's wrong, + of course. */ + current_buffer->enable_multibyte_characters = Qnil; coding.dst_multibyte = 0; }