Hans Verkuil <hverkuil@xxxxxxxxx> wrote on Fri [2019-Sep-13 15:34:28 +0200]: > On 9/13/19 3:32 PM, Benoit Parrot wrote: > > Hans Verkuil <hverkuil@xxxxxxxxx> wrote on Fri [2019-Sep-13 15:14:45 +0200]: > >> On 9/9/19 6:27 PM, Benoit Parrot wrote: > >>> Keep a reference to the currently selected struct vpfe_fmt * object. > >>> By doing so we rename the current struct v4l2_format * member from > >>> fmt to v_fmt. > >>> The added struct vpfe_fmt * reference is set to "const" so we also > >>> constify all accesses and related helper functions. > >> > >> This explains what you do, but not why you do it. > > > > Hmm ok I'll rework this a bit. > > > >> > >> And I think fmt vs v_fmt is *very* confusing naming. > > > > Well in this case v_fmt stands for "v4l2_fmt" and depending on the function > > I was using fmt to either mean a vpfe_fmt pointer or a v4l2_format pointer > > so tried (and apparently failed) to make it clearer which was which. > > Well, v_fmt could refer to either v4l2_format or vpfe_fmt, so that prefix > doesn't help me :-) > > Since 'fmt' was already defined in struct vpfe_device as v4l2_format, I'd > just keep that rather than causing code churn by changing it. And come up > with a better name for when you refer to a vpfe_fmt struct. Fair enough. > > Regards, > > Hans > > > > >> > >> I'd keep fmt as-is, and come up with a different name for the vpfe_fmt > >> pointer. ref_vpfe_fmt? > >> > >> Regards, > >> > >> Hans > >>
