Re: [openMSX-devel] Re: [openMSX-commits] CVS: openMSX/src MSXCPU.cc,1.20,1.21 MSXCPU.hh,1.14,1.15 MSXDiskRomPatch.cc,1.5,1.6 MSXDiskRomPatch.hh,1.4,1.5

[Joost Y. D. <j....@qa...>]

'Wouter Vermaelen' wrote about 'Re: [openMSX-devel] Re: [openMSX-commits] C=
VS: openMSX/src MSXCPU.cc,1.20,1.21 MSXCPU.hh,1.14,1.15 MSXDiskRomPatch.cc,=
1.5,1.6 MSXDiskRomPatch.hh,1.4,1.5' - Sun, Dec 30, 2001 at 09:29:13PM CET
>=20
>=20
> On Sun, 30 Dec 2001, Joost Yervante Damad wrote:
>=20
> > > No class needs to know there can be 2 CPU's in an MSX (turbo-R). Ther=
efore
> > > I suggest to remove the method getActiveCPU() and replace it with
> > > getCPURegs() and setCPURegs(). These 2 methods just pass the request =
to
> > > activeCPU.
> >
> > I also need access to readMem and writeMem, or should I do those
> > on the Motherboard? If soo I think your suggestion is excelent.
>=20
> You can access the MotherBoard directly. Or you can use the readMem()
> and writeMem() methods in CPU. The only difference is that the
> methods in CPU use the new caching mechanism. But disk is slow
> anyway, so caching in not important here.
>=20
> I'm not 100% sure caching works well (I mean in general, not
> specific for this case). I can't do an accurate performance
> measurment on my machine. Can someone with a slower machine check
> wheter the current caching routines are (much) faster/slower than
> without them?
>=20
> > > Maybe instead of 2 methods to get/set regs, 1 method
> > >       CPURegs& MSXCPU::getCPURegs()
> > > is easier? CPURegs& is not const thus you can also change the regs wi=
th
> > > this method. It is also slightly more efficient.
> >
> > As you wish ;)
>=20
> It's not my wish, it was just an idea. You are using these methods, you
> know what's best.

Hehe, I was joking, I meant: go ahead :) It seems
A Good Way to make the interface simpler.

BTW, disk support ALA fMSX is nearly finished.

Joost.