[PATCH 20/20] dix: reduce scope of tmp and mult.

[Daniel Stone]

Hi,

On Thu, Apr 21, 2011 at 11:58:03PM +0200, Simon Thum wrote:
> On 04/20/2011 10:21 AM, Daniel Stone wrote:
> > I'd also found and written 01/20 independently (it's sitting in another
> > branch) while having a similar 'wtf is all this?' excursion through the
> > pointer code, and had wondered how it worked!
> 
> Well, beyond a certain point there's simply no way I can make it better
> understandable for anyone but me in my current (at the time) mindset.
> Yes, it's complicated, but it's on par with the problem. And if I had
> gotten requests to clarify things in time, that might have been avoidable.
> 
> After all, the development process and review density surely improved.
> Today, some of peter's changes might have surfaced before even hitting
> master. But the real point is that if the issue is complex, the code
> will be too or it's too simple, meaning plain wrong.
> 
> I predict that X input will have to go some way towards more complexity
> to deliver the features one might naturally expect from a window system.
> There simply isn't much to argue about touchscreens bound to their
> output screen. X knows both. It's natural, it just requires some amount
> of in-server complexity - not some helper program floating around and
> fighting against other helper programs trying to achieve something else,
> piling up complexity on the system boundaries.
> 
> What can be argued about is reducible complexity, which means the scheme
> code here in accel. But this is the most straightforward part if you
> haven't been wading through mickeys before.
> 
> We can remove some of the settings and the scheme code without impairing
> functionality. We can make it simpler and better documented.
> 
> I'm happy to review patches.

Fair enough.  Thanks. :)

Cheers,
Daniel