Short:        PhxAss V4.37 68xxx Macro Assembler
Author:       frank@phoenix.owl.de (Frank Wille)
Uploader:     frank@phoenix.owl.de
Type:         dev/asm
Replaces:     dev/asm/PhxAss436*
Version:      4.37
Requires:     OS2.04

PhxAss V4.37 is a highly optimizing macro assembler for Motorola's 680x0
CPUs, 6888x FPUs and 68851 MMU (of course, the 030, 040 and 060 MMUs are
also supported). It requires OS2.04 (V37) and 1 MB RAM as minimum.

Most important features:
o PhxAss is FREEWARE.
o Symbolic and Source Level Debugging. Possibility to enable Source
  Level Debugging for high level languages too.
o Automatic generation of executables (if possible).
o Creates relocatable Amiga-DOS objects or absolute code (written into
  a file (raw or Motorola S-Format), into memory or directly onto disk).
o Small Code and Small Data support (also support for __MERGED sections).
o Listing file, Cross Reference Listing, Equates file.
o Complete floating point support: You can use complex floating point
  expressions, including float functions (sine, logarithm, square root,
  power, etc.), everywhere in your source, e.g. defining float EQUates or
  SETs.
o Switches for ten different optimizations.
o Locale symbols (xxx$ and .xxx type).
o Up to 36 macro parameters.
o Support for Motorola's old and new operand style (even in 68000 mode).
o locale.library usage (english, german, swedish, norwegian, italian,
  danish, dutch, hungarian and french ).
o Nearly all directives of the most popular assemblers like Seka, DevPac or
  AS (Aztec) are supported. Example: INCBIN, INCDIR, CODE_C, REPT, RS,
  RSRESET, EQUR, REG, OFFSET, XDEF, XREF, PUBLIC, OUTPUT ...
o PhxAss is able to stay resident in memory.

You will find two different versions of PhxAss in this distribution:
1. PhxAss: The standard 680x0,FPU,MMU macro assembler.
2. GigaPhxAss: Identical to PhxAss, but source codes are not limited to
   65535 lines. Quite useful for assembling huge Reassembler outputs.


CHANGES SINCE PHXASS V4.36 (last Aminet release):

Fixed Bugs:
o Optimization of CMPI.L #0,d(PC) into TST.L d(PC) generated a wrong
  displacement.
o Sporadic syntax errors on floating point expressions with exponent.
o Crashes after "word at odd address" in some situations.
o \+, \- and \. didn't work in strings.
o Severe problems with the EQU directive, when assigning an address or
  a distance from addresses from another section. In the first case
  wrong relocation information was written to the output file and in the
  second case the distance could get a wrong value. Example:
          code
  sym1    equ     sym2
          data
  sym2:   dc.l    0
o SHOWOPT always displayed "-56 bytes optimized" when optimizations in
  macros took place.

New Features:
o By using (s)pecial optimization, unnecessary branches like
    b<cc> label
  label:
  can be deleted.
o PhxAss is a bit more reassembler friendly now. ADD/SUB #x,Dn will no
  longer be converted into ADDI/SUBI #x,Dn.
o ADDI #0 wille also be optimized to TST (like SUBI #0 and CMPI #0).
o New directive SHOWOFFSET displays the current section offset via
  stdout (console).
o Source files which are completely empty, are allowed. PhxAss will
  generate an an object with a code section containing 0 bytes.
o CMPA #0,An is optimized into TST An, (n)ormal optimization and
  020 code generation are enabled.
o Some "68 Out of range" errors converted into "97 Immediate operand
  size error".
o New directives ENDASM/ASM to comment out multiple lines of a source
  text (comparable with IF 0/ENDIF).
o INITNEAR generates a single LEA base(PC),An if the small data
  section is identical with the current one.
o SF is optimized into CLR.B, if (s)pecial optization is enabled.