> > Linking wrapper functions is the better way IMHO, since it would not require
> > too much memory and let us run OpenBIOS in the background of these OSs.
> Yes, but in the early stages of development, it would be nice to have a working
> biosloader flashed on the chip, and then test new versions by loading them from
> disk. If we have a legacy bios loadable on disk, we can use that as a fallback
> then thing goes wrong.
Well, I'm currently working on a hardware solution, such as 512k RAM, writable from
another machine, possibly even with access to that machine's video RAM and keyboard
input (for remote testing... :-) ).
> This way we could also get started making some code. I
> don't know how much work this would take, but a simple thing that gets the first
> 128k from a floppy and then boots that would be a good start, wouldn't it?
Yes, of course, but how do you intend to test the BIOS loader code? I think this
code is much harder to write than the actual BIOS code, because of the memory
testing code which would clear the BIOS out of memory, I think. We would need some
kind of protected addtional memory.