[interfacekit] Re: :) B_MOUSE_MOVED

"Adi Oanca" <e2joseph@xxxxxxxxxx> wrote:
> > >     If you think 4 bytes will slow the system down... then, so be 
> > > it!
> > It won't slow down the system, but it'll be slower, that's a fact. 
> > Even
> > if it wouldn't be noticeable :)
>     Isn't that the point=3F

Almost, yes. But I used "noticeable", and not "measurable" - which 
might also be the case, but if it isn't, every cycle counts (yeah, 
okay, that's a bit exaggerated).

> > >     I don't think 4 bytes would make a difference, especially 
> > > that
> > > only one B=5FMOUSE=5FMOVED is in the system !
> >
> > It's not only slower than without it, IMHO it's just not necessary 
> > to
> > add it, because of the modifiers() function. Mouse movements are
> > generally handled ASAP, so the return value from modifiers() should
> > really be enough if someone is interested in that value (and even 
> > be
> > faster to call if it doesn't require to call the app=5Fserver).
>     How do you know that mouse movements are handled ASAP =3F I see no
> indication of that in disassembly code!

I don't know, but what I meant was the application point of view. I am 
not at all interested in what the mouse did a second ago, I do want to 
know what it is doing now.

> ...
>     I will do it as you guys say!, but I still remain convinced that 
> adding
> those 4 bytes will have no effect on performance!

Perhaps we should this thing on a list about things to consider for R2=3F 
It's really a subtle change, but once it's there, people might want to 
use it, and we couldn't even remove it again.
It's just that I can't find many use cases that would make this field a 
useful addition. And if there aren't any really convincing cases, I 
wouldn't add it to the API, no matter how big the speed impact would 
be.

Adios...
   Axel.

Other related posts: