Well, I'm confused, so I decided to turn to the august
community of erstwhile scholars we know as sys$mail:ee to answer
a question (or set of them). First a little background. When I
started working on ASAP, I needed names for data items of 8, 16,
and 32 bits. The first and last were relatively easy, 8 bits
was the minimum addressable unit, so it was a byte. 32 bits was
the size of all registers and the data bus so it was a word.
That means 16 bits would be a halfword. This was simple, self-
explanatory, and had historical precedent (IBM 360, PDP-10, etc).
I toyed with the idea of calling them char, short, and int, but
someone (I forget who) flamed me severely so I abandoned that
approach. So loading 16 bits was done with a LDH (sign extend)
or LDUH (zero-fill) while loading a whole machine word was LD.

	So far, so good. The fly in the ointment was that I was
assembling with a set of macros for MAC65, and .word generated
16-bit data while .long generated 32-bit data. I cobbled up a
version of MACAS to use .hword and .word respectively, but I
have never been _real_ happy with this and am getting less so.
When I ran the idea of reverting to the old definitions past
the two Eds, they agreed, but went on to say I should _also_
re-name the instructions. Under the proposed scheme we would have:

old		new		means
LD		LDL		load a 32-bit machine word
ST		STL		store a 32-bit machine word
LEA		LEAL		load the effective address of a 32-bit word
LDH		LDW		load a 16-bit halfword with sign extension
LDUH		LDUW		load a 16-bit halfword with zero-fill
STH		STW		store a 16-bit halfword
LEAH		LEAW		load the effective address of a 16-bit halfword
LDB		LDB		load an 8-bit byte with sign extension
LDUB		LDUB		load an 8-bit byte with zero-fill
STB		STB		store an 8-bit byte

	This leaves open the question of just what we should call the 32-bit
entity in comments, circuit diagrams, etc. It _is_ a word, but we have
pre-empted that (pardon the expression) word to mean something else. "long"
means something to C programmers, but not neccesarily to everyone. Also
there is the not-inconsiderable problem that my daughter's initials (LEA)
would no longer be a valid mnemonic :-). Other than then over-40 references
to words in the spec (and who knows how many in various programs) the proposed
scheme has merit, but I am still bothered. So I'd like each of you who cares
(or doesn't)  to reply telling me:

	"I prefer the proposed way as stated"
	"I prefer the proposed way, except forget the gratuitous 'L's"

	"I prefer the current way"
	"I prefer C (char) and S (short) and whoever disagrees can eat my..."
	"I don't care (and promise not to whine and snivel about final choice)"

   Those giving one of the first two answers should also be prepared to fill
in the blanks in the following sentence:

"You should call 32-bit quantities ______ in the hardware documentation"

	Of course none of this would be neccesary if we hadn't all stopped
learning after the PDP-11 or 8086, but as it confuses me too.....

Thank you for your support. Maybe next week we can get into how I could
have been so brain-damaged as to make the sucker little-endian, with a
latched carry, and no _____-straddling memory-references (I need that term
soon :-)
					Mike