[dpdk-dev] [PATCH v2 1/3] port: added WRITER_APPROACH == 1 implementation to ring port

[Thomas Monjalon]

Hi Cristian,

2015-05-19 22:49, Dumitrescu, Cristian:
> HI Thomas and David,
> 
> We can remove one of the code branches if you guys feel strongly about it.

We don't feel strongly as we are not authors nor testers of this new code.
But we want to avoid the maintenance nightmare of #ifdef.

> The reason we recommended to keep both is because both of them are working,
> and we wanted to keep both of them for a while to get some feedback from
> other people about which one they prefer. It might be that some apps would
> prefer one over the other for performance reasons. But again, we can resend
> this patch with only one code path present.

In general, RFC patches are used to request feedbacks.
I think it's better to have only one implementation at a time.
Comments are welcome.

> > -----Original Message-----
> > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Thomas Monjalon
> > Sent: Sunday, May 17, 2015 10:45 PM
> > To: Jastrzebski, MichalX K; Gajdzica, MaciejX T
> > Cc: dev at dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH v2 1/3] port: added WRITER_APPROACH == 1
> > implementation to ring port
> > 
> > 2015-04-30 13:58, Michal Jastrzebski:
> > > From: Maciej Gajdzica <maciejx.t.gajdzica at intel.com>
> > >
> > > Added better optimized implementation of tx_bulk for ring writer port
> > > based on
> > > similar solution in ethdev_writer port. New implementation sends burst
> > > without
> > > copying data to internal buffer if it is possible.
> > >
> > > Signed-off-by: Maciej Gajdzica <maciejx.t.gajdzica at intel.com>
> > [...]
> > > +#if RTE_PORT_RING_WRITER_APPROACH == 0
> > 
> > Nack
> > Maybe you missed the previous comment:
> > 	http://dpdk.org/ml/archives/dev/2015-March/015999.html