Re: An introduction to MI by example II (part B)...

Excerpts from mi400: 17-Nov-99 Re: An introduction to MI b..
Jon.Paris@halinfo.it (942*) 

> When you say [W-code is] "far above" [RISC instruction set], though what
> distance does that make the "move moon and terraform en-route" type
> instructions of MI ? 

"...in a galaxy far, far away..." *8-) 

It's always amazed me that, if the single-proton nucleus of a hydrogen
atom were enlarged to 0.5mm in diameter, that its electron would (on
average) be found 50 feet away...  So, if W-code is the nucleus of an
hydrogen atom...  No, wait...   If RISC instructions were subatomic
particles, then...  Ummm...  <sigh> 

You're absolutely correct that W-code is much "lower" than MI in many
respects, but the key difference is really the nature of the virtual
machine they model.  Comparing W-code's "stack-based" instruction set
with MI's "storage-to-storage" instruction set is a sort of "apples to
oranges" comparison, though the full function of MI can be (and is)
representable in W-code (with a few 400-specific swizzles). 

As you probably know, the original machine-level component that handled
program creation on the IMPI boxes was never even ported to RISC --
instead, a new component was crafted to convert the Original MI
instructions (in their 'program-creation template' form) into "MI-prime"
or "new MI" so they could be fed into the sole trusted code generator
component, the one that was originally crafted for ILE and the 'new MI'
in V2R3. 

Anyway, just some provocateuring...  I had to write lots and lots of MI
testcases "back when," so this is a nostalgic place to be.   

I certainly agree with the sentiments that MI is not really "close the
metal" -- the AS/400 doesn't usually let you get very close to the
metal, because it's hot and you might get burned.  :)   

Now, though, if you really want your logic to scream out loud, write it
in Java.  Java, with its pristine lack of ugly pointers, gets to exploit
features of the AS/400 that other HLLs can only dream of.  (Calls use
the machine stack, and pass parameters in <shudder> registers... 
Ooooo...)  Of course, the big hole in that argument is exactly tthe hole
that led us here, and that's providing interaction with legacy code. 
Java and legacy code just don't dance at the same sock-hop as well as
they might, yet.  However, writing OPM programs in MI isn't going to
break any land-speed records at run-time, either, as anyone versed in
late-bound program-call semantics is probably well-aware. 

Well, I've run out of hyphens -- time to restock. 

  ___   _           Blair Wyman                  IBM Rochester 
 ( /_)  /  _  ' _   (507)253-2891            blairw@us.ibm.com 
__/__)_/_<_/_/_/_'  Opinions expressed may not be those of IBM 



+---
| This is the MI Programmers Mailing List!
| To submit a new message, send your mail to MI400@midrange.com.
| To subscribe to this list send email to MI400-SUB@midrange.com.
| To unsubscribe from this list send email to MI400-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: dr2@cssas400.com
+---