Re: [PATCH] Support for case sensitive abbrev expansion

[Richard Stallman]

    The feature that makes it track `case-fold-search' probably isn't useful
    -- I expect that in most cases people will want to retain
    case-insensitive searching, even in situations where case-sensitive
    abbrevs would be appropriate.  I think you should just initialize the
    global value to `nil' for compatibility, and leave it to individual
    language modes to set a different buffer-local value.

Neither one is the right way to implement such a feature.  It would be
wrong for any mode to make all abbrev definitions case-sensitive.

All programming languages allow comments, which are written in human
languages, normally with mixed case.  Therefore, abbrevs for
human-language text are useful in all modes, and the current
case-adaptation features of abbrevs are useful in all modes.

The only kind of case-sensitive abbrev feature that would be
acceptable is one that allows individual abbrevs to be case-sensitive.
That feature would do no harm.  However, I don't think it is worth
implementing.  If you make your abbrevs three characters instead of
two, you can easily make them unobtrusive so that they don't get
in your way when you don't want them.