Re: address@hidden: skeleton bug]

[Richard Stallman]

    I found that this is because of the following incompatible
    change in emacs-unicode-2.

    Emacs-22: (char-or-string-p -1) => t
    Emacs-23: (char-or-string-p -1) => nil

    According to the function name, I think it should return nil
    on -1.

    Which do you think is better?

    (1) Revert the change of char-or-string-p.

    (2) Keep char-or-string-p as is in emacs-unicode-2,
        make a new function integer-or-string-p, and check all
        occurrences of char-or-string-p in *.el files.

I think we should keep the changed behavior of char-or-string-p,
which fits with its name, and check all 8 callers.