If you are so desperate to have it then write it yourself. Sheesh David McPaul > Cool Axel, well done Axel. Go on not doing stuff just for the fuck of > it Axel. > May you write something better Axel, and choke on it, Axel. > > Goddammit. You just lost another user. 2000 down, 3 left. Go Axel Go! > > ________________________________________ > On Monday 25 November 2002 17:38, you wrote: > > Eike Dehling <e.e.dehling@xxxxxxxxxxxxxxxxxx> wrote: > > > Ok. So it's at least a little more intelligent than some standard > > > blocklist > > > loader :) > > > > Sure - and it's very flexible, too :-) > > > > > Grub does have support for about 5 or 10 different filesystems, > > > so > > > there > > > are enough examples to look at when implementing BFS. > > > > Well, as I said, I won't do this. > > > > > Also it can load with blocklists instead of from a filesystem, so > > > it > > > could > > > already now load zbeos (maybe not pass parameters in the correct > > > way > > > ...) > > > > Theoretically, it could, yes. > > > > > But i don't see where your lack of real mode compilers comes from > > > ... > > > borlands turbo C (available for free from the borland museum), MS > > > quick C > > > (still commercial i guess) are just some examples. of course they > > > are > > > DOS > > > programs ... Although making it in assembly is probably easier > > > since > > > compilers tend to make rather big code, and 800 bytes is not very > > > much. > > > > Anyway, thanks for the hint, we might make some use out of it (I > > haven't spend time searching for them, yet :-). > > > > > >I will have a closer look at the multiboot standard before > > > > defining > > > > the > > > >protocol between stage 2 and the kernel. If it's not too much > > > > work, > > > > I > > > >would try to be compliant, just to not have to change the > > > > protocol > > > >again in the forseeable future. > > > > > > It is actually quite simple: your code is loaded in 32 bit mode > > > with > > > linear > > > 4gig segements and a20 enabled. > > > The code should be in a.out (requires a multiboot structure > > > somewhere > > > in > > > the code) or in elf. > > > You get passed a pointer to a structure with parameters in one of > > > the > > > registers. (memory map, command line, loaded modules, etc.) > > > > Well, as I said, I will have a look at it. If it is very similar to > > what we need and want to have, I will make it compliant, but I > > (myself) > > won't spend any time at improving GRUB to be able to load from a > > BFS > > partition. > > Which means that you'd have to copy the files needed to a special > > partition that GRUB can access, too, to boot with it, if you don't > > want > > to use our stage 2 boot loader. Anyway, being multiboot compliant > > (even > > if we are) won't mean that you can boot with GRUB the same way as > > with > > the native loader. BFS is one reason, the driver settings are > > another, > > and there might even be more. > > > > Adios... > > Axel. > Cheers David -- Philadelphia is not dull -- it just seems so because it is next to exciting Camden, New Jersey.