Re: Command line utilities argument types.

[Nurdin Premji <npremji at redhat dot com>]

Nurdin Premji wrote:
> Andrew Cagney wrote:
> > Nurdin Premji wrote:
> > > I'm going to be refactoring how the command line utilities will 
> > > parse their arguments to make them as unified as possible.
> > >
> > > I've identified 4 types of input that can be passed to a command 
> > > line utility: Pids, Core files, a command (executable with options), 
> > > or nothing.
> >
> > Would these be reported using a callback-interface?
> That is a great idea. I will use 3 callbacks to notify of pids, cores, 
> and commands.
> I'm thinking
> parsePid (ProcId pid)
> parseCore (File core)
> parseCommand (String[] command)
>
> I don't know whether the first two should return a single pid/core and 
> be called once for every pid/core found or arrays of pid/cores and 
> only be called once. The third callback would only ever be called once.
I have extended CommandLineParser in frysk.util to do the above, with 
callbacks parsePids(ProcId[] pids) parseCores(File[] coreFiles) and 
parseCommand(String[] command).

A CommandLineParser will call only one of these three callback 
functions. (There is no mixing of pids with cores withs commands.)