Re: two imenu entries for same function in C-mode

[martin rudalics]

> (On a related note, any regex junkies want to take a crack at the c-mode 
> regex(es)?  Some of those "identifiers" found in abbrev.c are buried in 
> comments.)
>
> -Dave
... you can find them in strings too.  However, regexps are not well
suited to establish whether an identifier is within a string or comment.
One could apply heuristics to rule out more blatant occurrences but this
might be expensive.  I see three ways to handle this problem:

(1) Impose strict rules on comment and string styles.  For example, you
    could require that any left paren within a comment or string must be
    preceded by a backslash.  However, such rules aren't even followed
    very closely by elisp programmers.  Also, routines for commenting
    out/in code would have to be rewritten accordingly.

(2) Rewrite Imenu in order to do some minimal parsing.  This would
    require to duplicate the first two passes of font-lock and write
    something similar to font-lock's third pass.  Suitable for people
    who don't like syntax highlighting.

(3) Let font-lock do the work.  In this case, the _entire_ buffer would
    have to be fontified immediately when visiting a file.  Stealth
    fontification could be only applied after buffer modifications and
    would have to be synchronized appropriately with Imenu lookup.