The static binary translators all output C code to run the emulations.

Currently the SBTs, when dumping a line of C, have to be told which
variables/flags/registers are read from or written to, in order to
remove dead code from the generated C.

That part of the coding could be simplified if the dead code elimination
were done separately, just acting on the C source directly without requiring
manual hints.

The plan had been to build a very simplified C parser that handled only
those statements generated by the SBT - and to that end we actually simplified
what was being generated in order to make it easier to parse; for example,
removing the need to handle operator precedence issues by insisting on
fully bracketed subexpressions.  And indeed a cut-down parser was written in
this style, which worked:

   http://gtoal.com/SBTPROJECT/6809sbt/experimental/no-precedence-parser/

(that one was simple enough that a quick hand-written parser sufficied)

However, writing this rekindled my enthusiasm for a previously-abandoned
project to write a C compiler, and it turned out to be relatively easy to
first generate a parser that accepted the full range of C expression
statements, and then incrementally add more constructs until it parsed
almost the entirety of the C language. (There may be a couple of very
obscure typedef constructs it still can't handle, but they are pathological
cases that only someone pathological would ever write).

The C parser is in:

   http://gtoal.com/SBTPROJECT/6809sbt/experimental/precedence-parser/

(Using a table-driven parser this time to make edits easier)

(You should also have a look at the notes in:

   http://gtoal.com/SBTPROJECT/README-HYBRID.txt

where a new possibility might make this whole area redundant.)