[Bruce Allen <ballen@gr...>]

Hi Phil,

> > ought to still be valid, but have long equivalents
> > 
> > -f: --autosave on
> > -F --autosave off
> 
> I like this suggestion, but we'd have to make a slight change: according to the
> glibc documentation, arguments to long options must be specified like this
> '--autosave=on'.

Yes, this is better.


> > as well as (perhaps)
> > -f: --autosaveon
> > -F: --autosaveoff
> 
> I think that one long option is quite enough ;-)

OK.  

> None of this should present any enormous problems, certainly not big
> enough that it should affect the decisions we make.

Good

> If only we could break backwards compatibility...
> I'd love to change options like:
> 
>        n      NotSCSI: Device is an ATA device.
>        N      NotATA: Device is a SCSI device.
> 
> Perhaps we could deprecate the worst and gradually remove them?

Let's make a good set of decisions about both the long and the short
command options, and then let's just break backwards compatibility
completely and do it right from now on.

In most cases, someone giving the "old" short options will end up getting
a usage message that will clue them in right away.

But let's spend at least a few days and decide on right right long and
short names for all the options first, OK?

Here's another technical question for you.  It might also be nice to
replace the Directive syntax for smartd.conf with a syntax that closely
matches the command line syntax for smartctl.  Is it possible to use
getopt_long to parse each line of the smartd.conf file?  In this case we
could use many of the same option names in smartd.conf as are used for
smartctl.  I *think* you can just set argc properly, construct your own
argv[] list, and then pass then to getopt_long, but there may be more to
it than that -- I am not sure.  Do you know?

Cheers,
	Bruce