On Fri, Jul 19, 2024 at 4:07 PM Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> wrote: > > Hi Andy, > > (CC'ing Mark) > > On Mon, Jun 10, 2024 at 07:31:11PM +0300, Andy Shevchenko wrote: > > On Mon, Jun 10, 2024 at 6:28 PM Laurent Pinchart wrote: > > > On Mon, Jun 10, 2024 at 06:06:51PM +0300, Andy Shevchenko wrote: > > > > Sat, Jun 08, 2024 at 05:16:33PM +0300, Laurent Pinchart kirjoitti: > > > > ... > > > > > Andy, we're reaching a level of nitpicking and yakshaving that even I > > > can't deal with. I will have to simply ignore the comments I disagree > > > with. > > > > Do you think using bulk APIs is nit-picking? > > In this case I do. If we were dealing with more 16-bit registers in this > driver I would agree with you. This being said, I'd like to get this > driver merged, and I'll burn some of the mental energy I've recovered > thanks to the last two weeks of holidays and submit a v5 using the bulk > API. It's getting mentally exhausting though. OK. > Overall, I think it would be nice to improve support for variable-length > register maps, in a similar way as done in include/media/v4l2-cci.h. > This driver, as well as many other drivers, could really benefit from > it. Mark, do you have an opinion, is v4l2-cci something that we could > fold in regmap itself ? +Cc: Hans as he might have even considered this (I'm speculating, but considering his quite a wide involvement in v4l2 sensor drivers and drivers that use regmap this idea might have come). -- With Best Regards, Andy Shevchenko