Open ColErr opened 12 years ago
Eh... just when the FS is finally looking up something like this shows up... I will try to take it into account once i'm done with high-level stuff.
The bootloader is good for the current gen of the spec now.
Sounds good. Though I have only a very faint understanding of what the issue was. In fact, I might even hold a wrong definition of what a bootloader is.
Right now I am assuming that it loads (the system) from a floppy disk into RAM.
Right?
Also: "Protects against partial reads."
does that mean that you HAVE to read one sector, 512 words, every time you run that interrupt? In short, you couldn't, for example, read just 16 words?
It protects against partial reads in that it won't read half the sector on one 'tick' and the other half on the next. The current spec only allows reading/writing one whole sector at a time.
The tick being referred to is a 1/60 second tick during which the game polls hardware.
Dammit notch, y u change specs?!
To be fair, he didn't change specs, we have just been using a 3rd party spec in the hopes Notch's would be similar.
Also, is there a spec sheet for the FS you are working on (or is it just FAT12/16) so perhaps we can help?
^ that.
this is the RFC the FS is based on https://gist.github.com/2782945 there were many changes, so if you have any questions, ask me.
http://pastebin.com/7wEKhdBx
It's not final, but he says it should be fairly close to production.
This means the FS and the bootloader will need some changes. I started trying to redo the boot loader, and it is almost working. The problem is the new spec will only read/write 1 sector at a time, and it would seem all operations are non-blocking, which makes the simple act of writing the first 0x10 sectors into memory a hassle. The one I have right now should work up until the first sector (which is written last). Easy fix, just need to throw a couple more lines to the end of memory.