Date: September 16, 1994
From: rodger@connected.com (Rodger Rosenbaum)
  To: jaykaye@kbbs.com (Joseph K. Horn)
  Re: <-RPL->                                                              

[Note: the following is just another reason to abandon the <-RPL->
library (on GD9) and use Jazz instead.  -jkh-]

Hi Joe,
    I've been trying to port Carlos Ferraro's POLY package to the GX,
and I have discovered a bug in <-RPL->.  When using ->COD to assemble
the following code fragment:

                   HS=0     3

I found that the assembler does not recognize the data field (the 3)
and assembles it as if it were:    HS=0   0.

     Both the SX and GX versions of <-RPL-> have the bug. 

---------
Note from Joe Horn:

It's worse than that.  Here's the scoop:

(1) The assembler should turn "HS=0 n" into opcode "82n", but instead
it always turns it into "820".  Workaround: use "NIBHEX 82n" as
recommended by SASM.DOC, or use "XM=0", "SB=0", "SR=0", "MP=0" or
"CLRHST" when n=1, 2, 4, 8, or F, respectively.

(2) The disassembler should turn opcode "82F" into "CLRHST" and then
increment the disassembly pointer by 3 to go to the next opcode.
Instead, it increments the pointer by 2, thus resuming disassembly at
the "F" of "82F" and thereby making a royal mess of at least one line
of disassembly after the CLRHST since the nibble boundaries of the
commands has been messed up.  Workaround: none.  Hint: After using
RPL-> with flag -4 set (disassembler turned on), perform << "CLRHST"
POS >> to see whether or not a CLRHST was disassembled.  If so, toss
the entire output of RPL->, turn off flag -4, and do the RPL-> again.

NOTE WELL: It is <-RPL-> that has this bug, NOT JAZZ. Jazz works fine.