[sheepdog] [PATCH v2] dog: permit two or more depth of subcommands

[Liu Yuan]

On Tue, Nov 12, 2013 at 04:33:49PM +0900, Hitoshi Mitake wrote:
> At Mon, 11 Nov 2013 21:24:47 +0800,
> Liu Yuan wrote:
> > 
> > On Mon, Nov 11, 2013 at 10:16:57PM +0900, Hitoshi Mitake wrote:
> > > From: Hitoshi Mitake <mitake.hitoshi at lab.ntt.co.jp>
> > > 
> > > Current dog command dies when user invokes "dog node log level set"
> > > without any arguments because of its subcommand design. This patch
> > > removes this limitation.
> > > 
> > > Signed-off-by: Hitoshi Mitake <mitake.hitoshi at lab.ntt.co.jp>
> > > ---
> > > 
> > > v2: define do_generic_subcommand() as a macro for a common case (depth == 0)
> > > 
> > >  dog/common.c | 5 +++--
> > >  dog/dog.h    | 5 ++++-
> > >  dog/node.c   | 3 ++-
> > >  3 files changed, 9 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/dog/common.c b/dog/common.c
> > > index 4480b99..90b43ad 100644
> > > --- a/dog/common.c
> > > +++ b/dog/common.c
> > > @@ -224,7 +224,8 @@ int send_light_req(const struct node_id *nid, struct sd_req *hdr)
> > >  	return 0;
> > >  }
> > >  
> > > -int do_generic_subcommand(struct subcommand *sub, int argc, char **argv)
> > > +int do_generic_subcommand(struct subcommand *sub, int depth,
> > > +			  int argc, char **argv)
> > >  {
> > 
> > Why we need 'int depth'? I don't think we need depth since we don't know where
> > is the horizon. What I meant in my last reply is just make 5 as a bigger one, like
> > 
> > #define DOG_MAX_PARAMETERS 10
> 
> What does the word "horizon" mean in this context? I think we need the
> depth variable for detecting the beginning of arguments for
> subcommands.
>

I meant 'beginning of arguements for subcommands'. But boundary of subcommands
depends on context. E.g, 'dog vdi log level list', vdi is subcommand of dog,
log is subcommand of vdi and so on. So I think we only need a marco to define
the max allowed subcommdans. An extra parameter isn't necessary to me.

Thanks
Yuan