[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] DATE/TIME cookies



On Fri, Feb 26, 1999 at 05:20:21PM +0100, Petr Stehlik wrote:
> > This is correct. But hopefully you agree that one should not introduce
> > services into the kernel which will BECOME costly once an already planned
> > change has been made?
> 
> Yes I agree (though see my estimation of slowdown in another mail).

It does not matter - why should we accept *any* slowdown if there is no
benefit, because the problem can be solved in a clean way without modifying
ans slowing down the kernel?

I did not get any example in this whole thread that could convince me why
the cookie would be really *necessary*.

> perhaps you mean the TSR doesn't use VDI for screen output. Or do you
> consider TOS' screen output (LineA) dirty, too?

You said you use direct screen access. Can you guarantee that this works on
any kind of graphics card? No, because it is not guaranteed that the screen
memory is memory-mapped at all.

As far as the TOS is concerned: TOS is specific to some kind of hardware,
and it knows about the hardware - if you want a different hardware, you need
a TOS that supports it. Since TOS is the operating system, it is allowed to
directly touch hardware, since that is exactly its task - in contrast to any
application program.

Another note: there is no separate LineA. LineA and VDI are strongly mixed
together - you can't clearly separate the two, at least in the TOS versions
I know. LineA is just an entry to some lowlevel routines in VDI, which may
or may not be there - VDI uses some of these, but it may also bypass LineA
in some respects, or use other routines which go directly to the hardware.

cu
Michael
-- 
Michael Schwingen, Ahornstrasse 36, 52074 Aachen