8086 memory space was: The Soul of Barb's New Machine 1154
Ok, let me go a little further. How are you going to load the code memory from the I-O device? You can't write to code memory using the DS-ES registers because they're pointing to memory somewhere else. The DMA controller (if you have one) doesn't know CS from an apple.
I didn't find 8086 buttembler terribly difficult to write. I did like PL-M quite a bit for a small language, but generally preferred buttembler. ...with a pot-full of macros. ;-)
I saw the same thing, but wasn't paid to do one and was well paid to do the other, so the decision wasn't especially difficult. The processor's ISA doesn't matter all that much to me. I even got to like the 8051's ISa (it was rather short on the 'a' part). ;-)
I didn't have any problems. It made sense to me. Sure, flat is easier, but segmentation wasn't all *that* complicated. Large data structures got to be a little unwieldy, but otherwise I didn't see any real problems with segmentation, per se.
"Huge" arrays-structures? I agree the huge model was a PITA, but as long as one stayed in the large model segmentation was rather painless.
8086 memory space was: The Soul of Barb's New Machine 1155
Indeed. It's a very old idea. I fear that it's been neglected too long because of the "cost" of implementation. Of course; spending an extra...
Alt Folklore Computers Newsgroups