Re: [dpdk-dev] [PATCH] ring: cleanup file-local macros at end-of-file

[Bruce Richardson <bruce.richardson@intel.com>]

On Wed, Mar 04, 2015 at 10:26:24AM +0000, Ananyev, Konstantin wrote:
> 
> 
> > -----Original Message-----
> > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Bruce Richardson
> > Sent: Wednesday, March 04, 2015 10:23 AM
> > To: Thomas Monjalon
> > Cc: dev@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH] ring: cleanup file-local macros at end-of-file
> > 
> > On Tue, Mar 03, 2015 at 10:03:45PM +0100, Thomas Monjalon wrote:
> > > 2015-03-03 16:38, Bruce Richardson:
> > > > The ENQUEUE_PTRS and DEQUEUE_PTRS macros defined in rte_ring.h are
> > > > not meant to be global and are not prefixed with the RTE_ prefix.
> > > > Therefore undef the macros at end of file to avoid pollution of the
> > > > global namespace, in case ends apps end up wanting to reuse those names.
> > > >
> > > > Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
> > > > ---
> > > >  lib/librte_ring/rte_ring.h | 4 ++++
> > > >  1 file changed, 4 insertions(+)
> > > >
> > > > diff --git a/lib/librte_ring/rte_ring.h b/lib/librte_ring/rte_ring.h
> > > > index bdf69b7..0d35648 100644
> > > > --- a/lib/librte_ring/rte_ring.h
> > > > +++ b/lib/librte_ring/rte_ring.h
> > > > @@ -1232,6 +1232,10 @@ rte_ring_dequeue_burst(struct rte_ring *r, void **obj_table, unsigned n)
> > > >  		return rte_ring_mc_dequeue_burst(r, obj_table, n);
> > > >  }
> > > >
> > > > +/* undef un-prefixed macros which are local to this file */
> > > > +#undef ENQUEUE_PTRS
> > > > +#undef DEQUEUE_PTRS
> > > > +
> > >
> > > Thanks for trying to clean-up things.
> > > Note that if an application is using this macro name, it will be destroyed
> > > when including rte_ring.h.
> > > Globally, DPDK namespace is awful and I hope we will be able to improve it.
> > >
> > Only if they are defining such a macro before including rte_ring.h, which I would
> > expect to be an edge case. Also, in such a case, the compiler/preprocessor will
> > give an error at the duplicate macro definition stage, and the simple fix is to
> > reorder the header file inclusion to avoid problems i.e. no changing of dpdk
> > required.
> > I suppose a better fix to go along with this is to RTE-prefix the macros. I'll
> > see about doing a V2.
> 
> Why not just to rename it to __RTE_RING_ENQUEUE_PTRS__ or something,
> and put in the comments that it is for internal usage?
> Konstantin

Yes, I'm planning on renaming it. But since it's for internal use only, I might
as well undef it at the end of the file too. I think it's good practice.

/Bruce