Guennadi Liakhovetski lg@denx.de wrote:
Oh great. We can do API changes without even mentioning it in the change log now?
Right, sorry, should have mentioned it. Although, the API change is one added function spi_select(), which you do not have to implement. So, I don't think I have broken anything.
The problem is that it is completely undocumented, and it appears to have a lot of overlap with the spi_setup() function I proposed several times a while back.
Besides, it's only optional up to the point where drivers start using it...and the mc13783-rtc driver does appear to be using it, so it isn't really optional anymore. And I suspect lots of other drivers really _do_ need this kind of thing, which is why I proposed the spi_setup() interface to begin with. You can always get around this by adding tweaks to the board/cpu/driver code for your particular setup, but I think very few drivers work as expected out of the box on a new board or platform. So I don't think an interface like this _should_ be optional.
Therefore, I'm going to remove it in the next version of my patchset. If you can tell me how it's supposed to work, I can try to minimize the breakage.
Haavard