> > > >> On Dec 10, 2018, at 4:06 AM, Jakub Grajciar <jgraj...@cisco.com> wrote: > > > > I do not like being the coding style police, but that is most of the > > comments > here and I will try to test this one later this week. Plus I am sure I missed > some > style problems, if you have not read the coding style for DPDK please have a > read. > > > > http://doc.dpdk.org/guides/contributing/coding_style.html > > > > One comment, why did you include all of the code to handle memif instead > of including the libmemif.a from VPP. I worry if libmemif is changed then we > have a breakage. I do not mind the PMD being standalone and I do like not > having the dependence. Just for my understanding, do you mean to say we could include the libmemif.a as a binary in DPDK?
IMO, I would like to view DPDK as the device abstraction and VPP as the protocol stack built on top. From this perspective, it is good to have standalone memif in DPDK. > > > > As I did not dive into the code much it does look reasonable and I hope to > give it a try later this week. > >> > > A couple more items, do you plan on writing the documentation for the PMD > and provide an example program? +1, would be good to have a cover letter. I would like to run this on Arm platforms, mostly in the beginning of Jan. > > Regards, > Keith
