Re: PATCH: Fix IDO interaction with uniquify.el

[Leo <sdl.web@googlemail.com>]

> What does exactly mean "kill virtual buffers"? If you mean removing
> them from the buffer list, are you going to maintain a separate list,
> instead of regenerating it from recentf-list anew for each
> ido-switch-buffer, or are you going to modify recentf-list?

'kill virtual buffers' means removing them from recentf-list i.e. they
cease to be virtual buffers. So the latter.

> As for the duplicate entries, have you implemented it without using
> uniquify? That does not seem too hard, and it's definitely cleaner.

John and I have discussed whether uniquify should be used. He wants
the virtual buffer name to follow the same style as that from
uniquify. But in order to use unquify, some changes need to be made
since at the moment uniquify is buffer oriented and virtual buffers
are not buffers. Secondly to uniquify completely, the full list of
files with same base name must be known and this adds complexity. I
locally have a patch that completely handles duplicate basenames
(about 40 lines of elisp without using caching) without using uniquify
so I have some idea about the complexity.

In the end we decide temporarily just adding some number
(customisable) of parent directory. In practice one level of parent
directory already significantly removes the chance of a file in
recentf-list being ignored.

>    Juanma

Leo