#: 8942 S12/OS9/68000 (OSK)
    29-Dec-90  20:07:31
Sb: #8928-68000 ASM Language
Fm: Jack Crenshaw 72325,1327
To: Ed Gresick 76576,3312 (X)

Ed, the main reason for wanting to do an assembler is to get relocatable code
... the current SK*DOS assembler only does absolute.  It's ASM from Bud Pass. 
So the linker is a definite must.

Right after the PT-68 came out, Sidney Thompson recruited me to build a linker.
He and Bud Pass were developing a C compiler, but with no linker it makes
building modular software pretty difficult!

I did build the linker, but we had a chicken-and-egg problem:  A great linker
is not much good if the assembler and compiler can't generate the object
modules for it!  Really, the linker and assembler have to be built as parts of
a system.  So I had to begin with a linker format that ASM was capable of
generating, i.e. I was limited to what I could inject into the object file via
DC statements.

The linker's been done for about two years now, but Sidney and Bud don't like
it (not elegant enough for Bud, though he declined to offer any suggestions as
to improvement).  Neither has shown any interest in incorporating it into the
assembler or compiler,and Bud's support of SK*DOS seems to have dropped off to
zero.  So I figure if we're ever going to get a good relocatable
assembler/linker/compiler system ... in other words, adequate development
tools, I'm going to end up writing them.

Which brings us back full circle.  In a few attempts at prototyping the
assembler, I realized I was spending a lot of energy to handle a syntax that is
fundamentally flawed.  It seemed to me that, as long as I'm gonna do this,
might as well make a few improvements along the way.

Jack