Re: address@hidden: change to file on disk not noticed by indirect buffer]

[Chong Yidong]

Richard Stallman <address@hidden> writes:

> The failure to ask the user whether to update the buffer
> is certainly a bug.  Would someone please fix it and ack?

I fixed the clash detection bug.  (I didn't do anything about the
other issue with find-alternate file.)

> The behavior of find-alternate-file is also surely not good, but I am
> not sure what it should do instead.  The ideal would be to substitute
> the new file in the base buffer and its indirect buffers, without
> otherwise altering the differences between them.  But that is probably
> quite hard, so it would be better to do something else that's at least
> sensible.
>
> From: Joe Edmonds <address@hidden>
> Subject: change to file on disk not noticed by indirect buffer
> To: address@hidden
>
> I really like using indirect buffers to edit different parts of the
> same file with different buffer settings (mode, point, mark, position
> in buffer stack, etc).
>
> But I've noticed that an indirect buffer doesn't behave the way a
> normal buffer does with regard to a file that is changed from outside
> emacs.  Here's how to reproduce the problem:
>
>   echo foo >>/tmp/foo   # in a shell outside emacs
>
>   C-x C-f /tmp/foo
>   M-x clone-indirect-buffer
>
>   echo bar >>/tmp/foo   # in a shell outside emacs
>
>   Now, insert some text in the indirect buffer.
>
> You don't get the nice "foo changed on disk; really edit the buffer?"
> prompt emacs normally gives you if you try to edit the base buffer.
>
> Also, find-alternate-file destroys the correspondence of indirect
> buffers and base buffers.  If you do it in a base buffer, all indirect
> buffers are killed.  If you do it in an indirect buffer, you're
> not an indirect buffer anymore.  If there were a "refresh base buffer
> and all its indirect buffers from disk" function, that would make
> working with indirect buffers easier for me.