On 3 February 2015 at 18:47, Taras Kondratiuk <taras.kondrat...@linaro.org> wrote:
> On 02/04/2015 12:42 AM, Mike Holmes wrote: > > > > > > On 3 February 2015 at 17:16, Taras Kondratiuk > > <taras.kondrat...@linaro.org <mailto:taras.kondrat...@linaro.org>> > wrote: > > > > On 02/03/2015 11:50 PM, Bill Fischofer wrote: > > > The changes here are to > > > platform/linux-generic/include/odp/plat/<type>_types.h. This is > where > > > the strong typing is actually implemented for each ODP type in > > > linux-generic. The converter functions are part of that > implementation, > > > and are also inlined for performance, which is why they are here. > > > > > > All other changes within each patch part are the modifications to > other > > > ODP files needed as a result of switching to strong typing for this > > > type. They are not separable since once strong typing is enabled > for > > > that type the code won't compile without these changes. > > > > > > The doxygen stuff is in the APIs which are in include/odp/api/*.h > as you > > > mention. The main difference between v4 and v5 was to break out > the API > > > definition changes separately, which is why this has gone from 10 > to 19 > > > parts. > > > > > > So it sounds like this is doing exactly what you suggest, no? > > > > As I've commented in the previous message it is OK, because these API > > are kind of exceptional. > > > > > > I have no issue making an exception, but I don't yet grasp the reason > > why we make it, what do we gain by doing so ? > > If some module need to use odp_buffer_t it is going to include > odp/plat/buffer_types.h. This module may need to print odp_buffer_t for > debug > purposes, so odp_buffer_to_u64() should be accessible. > Thank you -- *Mike Holmes* Linaro Sr Technical Manager LNG - ODP
_______________________________________________ lng-odp mailing list lng-odp@lists.linaro.org http://lists.linaro.org/mailman/listinfo/lng-odp
