how to add support for more Unicode characters?

[Hrvoje Niksic]

David Kastrup <dak at gnu.org> writes:

> GNU Emacs is rather rigorous with line endings: have a single
> inconsistent line ending in the buffer, and the line ending mode is
> thrown into Unix.  Thus line endings are correctly reproduced, too.

This is only guaranteed if the file isn't modified at all.  (This has
just occurred to me while reading Aidan't reply, sorry about the late
response.)

If a binary file happens to contain one or more CRLF sequences without
any instances of single CR or LF, it will be read in "DOS" line-ending
mode, right?  In that case, the user who goes to end of buffer and
presses `C-q C-j' thinks he is appending a single byte to the file,
when he is in fact appending two.  Likewise, the user who replaces
every instance of "a" with "C-j" might think that he's not changing
the length of the file, whereas in reality he is (and at that
silently).  So unfortunately it's not completely true that opening
binary files can't corrupt them as long as you stick to ASCII.

While I admit that the above situation is unlikely, it is entirely
possible.  Unless I'm missing something, one cannot claim that GNU
Emacs always correctly reproduces the file contents -- except when the
file is not modified at all.

This is not a criticism of GNU Emacs's behavior -- I still consider it
superior to what XEmacs does and would like to learn from your
experience.  But in general case it appears impossible to have *both*
seamless editing of binary files and seamless decoding of UTF-8 and
EOL sequences in files at the same time.  (By seamless I mean "without
interrogating the user.")