> Patrik Stridvall <[EMAIL PROTECTED]> writes:
>
> > The device code has four choice: Provide a function
> > taking logical coordinates or a functions taking
> > device coordinates or provide both or provide none.
>
> Then when you read the device code you have no idea what kind of
> coordinates the function will receive unless you go hunt the
> corresponding function table. I definitely think it makes the code
> less readable.
As the code is now, yes. However that has nothing to do with
the infrastructure itself.
Most driver will have functions that are almost exlusively
of one type, device or logical, so it not really a problem.
So the few exception should be marked somehow like
void FOODRV_DEV_Bar(void)
or
void FOODRV_LOG_Bar(void)
Sure, if we take the most important example the X11 driver
all functions currently take logical coordinates and
they are supposed to be converted one by one to
device coordinates. When each functions is converted
it is renamed to X11DRV_DEV_<function> to mark this.
Whether this is temporary or permanent is an entirely
different issue.
In the long run we have basicly two choice
1. All functions are prefixed "DEV" or "LOG".
2. Only the exceptions from the normal case
of the device are is marked.
In short, it is only a tempory problem and a small
one at that. For example I can easily make a patch
that converts the current X11 driver function to
X11DRV_LOG_<function> so the device version can
be X11DRV_<function>. There are also multiple other
possibilities. Pick one and I will do it it is of
no consequence what so ever.
I understand that you are worried but this is
not something to be worried about. As I said
every device is (or rather will be) almost
exclusivly of one type and the exceptions
can be clearly marked. Transitional issues
are easily solvable as well as meantioned
above.
