TO DO list for exim_help

*   There is no way to get the patterns from a file
    yet.

*   There is no way to combine patterns in a boolean
    manner.  Any that match result in output
    although all the patterns can be ANDed 
    together with -a.

*   Given the number of options (line numbers, 
    case sensitivity, HTML output, ...) a GUI 
    for this might be useful.

*   It seems possible to hold all the information
    about paragraph boundaries, numbers of words
    in a pargraph, etc, in a database.  This would
    be done on a per file basis, and if the database
    were newer than the file, it would be read,
    otherwise ti would be cretaed.  This would speed
    up subsequent help requests on the same file, 
    and would simply be rebuilt when the newer doc
    files were installed.   This is subject to
    demaned.

*   The determination of whether a paragraph is a
    descrion of a variable (option) could be makde
    more sophisticated, by makeing sure that the
    paragraph holding the option name is not only
    one word, but matches /^\$\w+\:?/, but this 
    would slow things down.  Are the false positive
    matches annoying enough to merit this?

