ABI questions

[Peter Hutterer]

On Tue, Oct 19, 2010 at 03:48:48PM -0400, Trevor Woerner wrote:
> On Tue, Oct 19, 2010 at 3:42 PM, Thomas Hellstrom <thellstrom at vmware.com> wrote:
> > The Sep 18 InputDriverRec loses refcount hopefully occured on the master
> > branch, and *should* have been accompanied by an XINPUT ABI bump to 12,0,
> > but apparently somebody missed that.
> 
> So that explains why the evdev device checks for XINPUT ABI 12 or
> better even though the XINPUT ABI is only at 11 (that was going to be
> my next question) :-)
> 

that's because I have an input-api branch that changes large parts in my
tree but I'm waiting for two more changes. that branch bumps the abi to 12.
i was ready to merge it a while ago when I pushed the matching changes to
the input drivers, but then had to revoke the pull request because a few
bits were missing. that's why the drivers already check for abi 12.

as for the refcount change - that should have gone into the same branch but
it ended up in master when I wasn't paying enough attention. it's harmless
though since the refcount hasn't been used by anything for ages, so a simple
abi > 11 check to remove it would do. for the few git versions in between,
all you get is a warning.

Cheers,
  Peter