Re: z80 C cross compiler/backend

wilker@hopf.math.purdue.edu (Clarence Wilkerson) wrote:


>There was also Software Toolworks C80-3.0, A Digital Research C compiler,
> and Mix C. The latter was sold until fairly recently.


tsullivan@serv2.fwi.com (Tom Sullivan) writes:
>And yet another was the Eco-C compiler. I had that one, and it
>generates Z-80 assembler as its output, even came packaged with the
>M-80 assembler and linker for assembling its output.
>
> Tom Sullivan tsullivan@mail.fwi.com


The later versions of Eco-C came with a stripped down version of the
SLR Z80 assembler and linker instead. This apparently let EcoSoft cut
the price to around $50 compared to $200 when they were shipping M80
and L80.


The big thing that I ran into when I was using SA80 was that it didn't
have macros! I modified the source for the startup code to save the
HL register so that C programs could access the Z-System environment.
The problem was that this source used macros, which the SA80 assembler
couldn't handle, to define the space used by the file tables (it
allows for 6 files by default)! So, I just calculated the number of
bytes needed and used an appropriate 'ds' pseudo-op to set up the
table.


I used Bridger Mitchell's article from the Computer Journal as a basis
for the Z80 code to verify that I had a good Z environment address. I
gave up on the project for two reasons: the primary one was that
Eco-C's compiler would run out of memory whenever I tried to compile
something with the header files I wrote to describe the Z environment
-- I had to drop back to plain CP/M to compile anything useful; the
second reason was the very slow and inefficient code that this
compiler produces.


Has anyone tried to Z-ify the Hitech C compiler? It generates much
better code than Eco-C (speed-wise it is on a par to Turbo Modula 2),
and it is more UNIX-like than BDS-C.
--
Jeff Wieland
Jeff_Wieland@ces.purdue.edu
--

Return to the comp.compilers page.
Search the comp.compilers archives again.