I managed to contact the original create of the program, Clinton, and he
tracked it down to a change in Rex:
"As expected the error was caused by the location of the directory in REX
having changed. Rebuilding with the correct address ('namdir equ 06100h' in
the asm file) makes it work with REX 4.9 (tested in VirtualT.)
If you decide to rebuild and use it, PLEASE back up anything important from
your RAM first, and regularly. There are some things I came across in my code
that I want to verify will work safely, and fix if needed. “
I made the change and am now able to play Zork.
> On Dec 3, 2025, at 6:26 PM, Joshua O'Keefe <[email protected]> wrote:
>
>> On Dec 3, 2025, at 1:48 PM, Mike Hyde <[email protected]> wrote:
>>
>> Could the changes in REX from 4.8 to 4.9 be the cause?
>
> I built the tool with the recommended namdir at 06400h, for REX 4.9 (which is
> what I have), but I tested in virtual-t, which has a 4.8 REX by default. I'll
> try again.
>
> [...]
>
> It works as intended on virtual-t when z100.hex is built with namdir equ
> 04400h (which was the default). That leads me to expect a REX 4.9 should work
> fine on real, modern hardware. I don't have the runtime environment to build
> a Windows binary but when the include is a 4.8 include and z100pkg is rebuilt
> (at least on macOS) using that, it runs on virtual-t.
>
> It might be helpful if a Windows-y person could come along and build the
> Windows binary with the reassembled loader. That should work how you expect
> it to.
