On 17/08/15 17:21, Andrew Lunn wrote: >>> If you think your PWM code is so much better, please submit a revert >>> patch plus your PWM LED driver. We can find somebody to do a side by >>> side review. >> >> It's not about the driver code. That can be cleaned up if needed. The >> question is whether the driver should be a PWM driver or a LED driver. > > There i disagree. It is all about the code. Release early, release > often. Show us the code. Get it discussed, reviewed, tested, and then > ultimately merged. > > You keep saying PWM is the way to solve this problem, but where is > your code showing your solution is superior to mine, yet still solves > my use case? No, I have not said PWM is the way to solve this problem. I have said it sounds to me that a PWM driver better matches the HW and our use case. I have asked you if PWM + pwd-leds would solve your use case as well as a led driver. If you don't have an answer, it's fine to say "I don't know". But you have just ignored the question. And even worse, you dropped me (and Vignesh) from the latest mail thread. We have a pwm driver for the chip, but it only supports TLC59108. It should be trivial to extend it to support TLC59116, but it's pointless to spend time on that if a PWM driver doesn't handle your use cases. > So please stop talking and show us the code. TLC591xx is quite simple HW, and the code doing the HW programming should be identical for a pwm and for a led driver. I don't see either a pwm or a led driver being somehow superior. They are very similar simple drivers, implementing a different interface. Whether the interface works for you should be clear from the interface itself. The code is just details. On the other hand, if you want something you can use to test, it's a different case, but you haven't indicated any interest in that. Tomi
Attachment:
signature.asc
Description: OpenPGP digital signature
