On 8/10/19 9:05 PM, Dmitry Torokhov wrote: > On Sat, Aug 10, 2019 at 08:00:14PM +0200, Marek Vasut wrote: >> On 8/10/19 7:44 PM, Dmitry Torokhov wrote: >>> On Sat, Aug 10, 2019 at 06:50:08PM +0200, Marek Vasut wrote: >>>> On 8/10/19 6:41 PM, Dmitry Torokhov wrote: >>>>> Hi Marek, >>>> >>>> Hi, >>>> >>>>> On Sat, Mar 02, 2019 at 03:17:04PM +0100, Marek Vasut wrote: >>>>>> Add support for ILI2117 touch controller. This controller is similar >>>>>> to the ILI210x and ILI251x, except for the following differences: >>>>>> - Reading out of touch data must happen at most 300 mS after the >>>>>> interrupt line was asserted. No command must be sent, the data >>>>>> are returned upon pure I2C read of 43 bytes long. >>>>>> - Supports 10 simultaneous touch inputs. >>>>>> - Touch data format is slightly different. >>>>> >>>>> So with this and also I see there is another ili2117a submission, I do >>>>> believe that we need to switch to using function pointers instead of >>>>> if/else if/else style cheking of the model. >>>> >>>> How about we add tested functionality in first and only then do bigger >>>> untested changes ? I think that would work better for everyone. >>> >>> Sorry, I would really prefer to do what is right and build additional >>> functionality on top of the good foundation. I asked to switch to the >>> function pointers before, but you did not want to citing performance >>> (even though we are using function pointers everywhere in the kernel), >>> now I gave a draft implementation, I hope you can use it. >> >> So why can't we add tested code in first and then add new huge untested >> patch on top and start testing it ? I think doing it in reverse is >> actually not helpful, if there is a problem in this massive new patch, >> it could be reverted without losing functionality. > > We still have 4 weeks till merge window + stabilization time past it. Sure, but this patch was posted 5 months ago and was in real world deployment since, so it has 5 months of practical testing. I don't want to throw that away. The patch you want me to test can easily be rebased on the ILI2117 support and then we retain those months of testing, which I think is much better.
