Slinky is getting wrong answers when writing the binary

[michaelcmartin]

If I create the hello.m test program and use the commands you state on your blog to build it, the linker fails as it attempts to jam the value $3ff8 into a single byte. On Linux, the link succeeds but the value loaded in the binary for the "text" variable is $0705 instead of $c007. Oddly enough, this only appears in the .bin file; the .sym file correctly names the intended locations.

[michaelcmartin]

Okay, I've figured this out. The object-code parser relies on a y.tab.h and it has to be the same y.tab.h as the one generated during the macross build. Otherwise evaluation of expressions goes wrong.

A proper fix to this is to reorganize the Makefiles so that macross and slinky are both sharing the header file as appropriate.

[michaelcmartin]

I don't have a whole lot of experience with build work, but I think I've been able to strip it down to its minimum reasonable ruleset in the commit above. I don't know if it ends up restricting compatibility any, but I'd hope it doesn't.