- Relocation sections--I had prototyped an implementation but removed it because
the implementatiof of the symbol table in
ELFwas wrong. There should be a relocation section for every other section (e.g. ".reltext").
- Lots more directives.
- More instruction types--currently partial support only.
There are already lots of test programs and the expected tokenizer output for each one. For at least a subset of these programs, develop a method for providing them to the parser as input and checking the resulting AST. Then, check that the object file produced by the assembler is correct as well.
For the parser, this could mean writing code to load an AST represented by
s-expressions into a tree of
Nodes. The code already exists in
assembler_test.cpp to compare ASTs.
For the assembler, I guess this means constructing some ELF files by hand. The more flexible approach would be to define some notion of equality between two ELF files so that two sections, each defining the same symbols but in a different order, are considered "the same." The simpler approach would be a bytewise comparison of the two files.