> > > I'm ok with -l, -d as well, but another option is to follow the
> > > getsubopt(3C) model of -a local=<addr>[,remote=<addr>].
> >
> > I like that much better, yes. One would then have the option of setting
> > the remote address independently by using "-a remote=<addr>". Quite
> > nice.
>
> On a related note, what is the correct way to represent the above syntax
> in documentation and usage synopses? Consider the following possibly
> incorrect create-iptun syntax:
>
> dladm create-iptun [-t] -T <type> [-a local=<laddr>,remote=<raddr>] <link>
>
> It's possibly incorrect because when -a is specified, either local or
> remote or both need to be included. So they're simultaneously optional
> and not optional...
Are suboptions usually included in the synopsis? I thought we usually
just say something vague like "[-a <addrtype>=<addr>,... ]". If we did try
to wedge it in there, I'd think it'd be something like:
dladm create-iptun [-t] -T <type> [-a {local|remote}=<addr>,...] <link>
--
meem
