Re: How does Emacs find the buffer for a file?

[Richard Stallman <rms@gnu.org>]

    When you do C-x C-f /some/file RET, then Emacs runs
    insert-file-contents and puts the output in some buffer.  How does
    Emacs determine the buffer from the filename?

Actually it creates a buffer and then runs insert-file-contents.  So
this question does not really make sense.  Emacs does not determine a
buffer, it uses the buffer it just created.

The code is in find-file-noselect and find-file-noselect-1.

    There is a bug in Tramp, and I don't know how to start tracking it
    down.  Tramp allows you to edit remote files.  If you say C-x C-f
    /foo RET (editing the local file /foo), and then say C-x C-f
    /[user@host]/foo RET (editing a remote file, also named /foo, via
    Tramp), then the remote file contents are inserted into the buffer
    containing the local file /foo.

Assuming Tramp uses the file-name-handler-alist mechanism, the only
operation that it will handle is insert-file-contents.  That handler
should insert into the current buffer.  It should never try to find
the buffer to use.

Basically, you need to make sure things are working with Tramp more or
less the same way they are working for local files, aside from
fetching the data of course.  A problem like this would tend to come
from a difference where the two paths ought to work the same way.