[Libva] [PATCH v2 3/5] API: backend: rename dri_state to drm_state.

[Gwenole Beauchesne]

Hi,

2012/7/9 ykzhao <yakui.zhao at intel.com>:
> On Sun, 2012-07-08 at 23:06 -0600, Gwenole Beauchesne wrote:
>> 2012/7/9 ykzhao <yakui.zhao at intel.com>:
>> > On Sun, 2012-07-08 at 22:28 -0600, Gwenole Beauchesne wrote:
>> >> > There is no problem when the variable of dri_state is renamed to
>> >> > drm_state.
>> >> > But it seems a little confusing as the patch 2 already defines the
>> >> > structure type of drm_state.
>> >>
>> >> Both patches could be merged altogether, if this is what you suggest.
>> >
>> > No. What I mean is that the name seems a little confusing. And the
>> > readability of ctx->drm_state is decreased as the patch 2 defines the
>> > data structure of drm_state.
>>
>> What are you really talking about? And what's your suggestion then?
>
> It is OK for me to rename the "dri_state" to drm_state. But to keep the
> readability of "ctx->dri_state/drm_state", I suggest that the drm_state
> defined in patch 2 can use another name. (For example: drm_base_state)

I am sorry but I still don't see your point and how this affects
readability. drm_state is the base class for dri_state, which is a
DRI-authenticated drm_state + data for its housekeeping. It's pretty
clear already. Your suggestion actually decreases readability when it
comes to the VA/DRM case whereby you don't care of extra data, so the
base drm_state is already sufficient.