At Tue, 28 Apr 2009 15:59:52 -0400, Jon Smirl wrote:
On Tue, Apr 28, 2009 at 3:53 PM, Mark Brown broonie@sirena.org.uk wrote:
On Tue, Apr 28, 2009 at 09:37:34PM +0200, Takashi Iwai wrote:
Also, we need to reconsider which of you change is really needed since the current problem of the PCM will be solved in the PCM core side soon later without the change of the driver side.
AFAICT all the currently posted patches are needed since they're only doing the refactoring of the code required to support AC97. They're only related to the fixes in that John's board doesn't have I2S but since the DMA is shared the fixes that John develops while making AC97 work will also fix I2S.
I purposely sent in refactoring changes that made no functional changes to the code.
I have been caught in this mess before. This isn't a simple case of resolving conflicts.
Why do you believe that this case will be like that, too?
What happens is that git isn't smart enough to track changes across a refactor. That results in big conflicts covering most of the contents of the files involved. The conflicts in the refactor then cascade into all of the other patches.
If we put the refactor in front of the fixes git will get everything right.
Well, you can do it by yourself on your local tree, too. topgit or other tools can serve well for such use cases. Then either let Mark or me pull your tree, or straighten the patches to submit.
Why are people going to complain about patches to a driver marked broken? You can't even compile it without editing the Kconfig.
Because we broke it.
Takashi