On Sat, 10 May 2008 03:21:35 +0100 (BST), Maciej W. Rozycki wrote:
> > This was an option when the functions where introduced 9 years ago.
> > But now that it was done, renaming them would cause even more
> > confusion, I think. I would be fine with adding comments in i2c-core.c
> > or improving Documentation/i2c/smbus-protocol to make it more obious,
> > though.
> > On a related note, you will notice that the other i2c_smbus_* functions
> > do not follow the naming of SMBus transactions. Again that's something
> > I regret but I feel that changing the names now would cause a lot of
> > confusion amongst developers, so I'm not doing it.
> It may not be worth the effort, but if done in bulk for all the users in
> the tree, there should be no problem with that. I am fairly sure there
> were changes of this kind from time to time, with occasional screams heard
> in response from some dark corners, but no big pain. We obviously
> explicitly disregard out-of-tree users and for occasional contributors
> asking: "Where the * has this function gone?" there is the Documentation/
> tree to provide a greppable reference, so generally not a big deal.
It's not that easy. There are some drivers which are both in-tree and
out-of-tree, for which such a change means adding ifdefs. And there is
i2c-dev.h (the user-space one) which has similar functions, if we
rename only the kernel variants, there will be some confusion. But if
we rename also the user-space variants, then it's up to 2.4 kernel
users to have different names for kernel-space and user-space functions.
All in all I'd say it is not worth the effort. There are many other
tasks where our time will be better used.
> > Just one patch should be enough, if I agree with all the changes. You
> > might make a separate patch with the things I may not agree with, so
> > that you don't have to cherry-revert them if I indeed don't agree, and
> > we just merge them if I do agree.
> Hmm, technically you do not seem to be responsible to accept changes
> under drivers/rtc/, so I will split them anyway for others to decide.
Huu, sorry, for some reason I thought that we were still speaking about
i2c-sibyte. Of course I don't have my say about what happens in