Sat Mar 19 10:56:51 EDT 2011

Weird bug

Figuring out the backtraces seems a bit much work.  I used a tracing
approach, tagging the macro invokations with a `printf'.  Traced it
down to:

    UEP0 >block


   : >block 16 / ;

My guess is that UEP0 is undefined.

No it's something more insiduous.  In ramblock.ss, the following code
is compiled as if `>block' is a target word, not a macro.

: >block 16 / ;

Changing the `:' to an explicit `:macro' solves the problem.

Ok, I moved the code that was in coma/macro-forth-tx.ss back to a
`begin-for-syntax' form and the problem goes away.

I don't understand..  Something to do with local state maybe?

Bottom line: that code has to change.  Too obscure.