Ian Collier wrote: > It was basically a choice between keeping 8-bit quantities and > shifting+ORing whenever register pairs are used, and keeping 16-bit > quantities and shifting whenever single registers are used. I made the > choice arbitrarily as it seemed easier to deal with the 8-bit registers > individually.
I was thinking of a packed structure (but not necessarily in a strcuture) so that B is stored in the the location following C, so you can pick each out individually or BC as a WORD, without needing to shift or OR anything. Picking the word out was the bit that I was thinking would be endian sensitive. It'd probably work if done as a C union, but I wasn't sure how efficient that'd compile up to be compared to the simple variable version. I might give it a try with HL to see what different it makes. Si
