An haskell clone of the original java Gom code generator.
To compile hgom, install stack. Then run:
hgom binary is generated in a directory displayed by stack. You can run
it from there. Alternatively you can run it using
stack exec hgom.
hgom --help to get some basic help.
You can test the behaviour of
hgom by running it as follows. Some examples
are valid files, other ones demonstrate
hgom error messages.
hgom examples/simple.gom hgom examples/big.gom hgom examples/many_errors.gom ...
Run the tests with:
stack test --test-arguments "-a N"
N is the number of generated random inputs for each test case.
For a list of all possible test arguments check
stack test --test-arguments "--test".
There is some benchmark in
test/bench that generates bigger and bigger gom
files and runs
gom on them, measuring the gom/hgom ratio
concerning the number of generated lines (using
sloccount) and the elapsed time.
cd bench make
It takes some time. The generated files can be plotted using gnuplot for instance.
Differences with gom
- a far less permissive checker
- smaller code, compiles much faster
- almost 100% code coverage,
- unit tests using QuickCheck, both on compiler data structures and generated code
- regression tests for parser and checker
- more things optional: visitable, checker, ...
sizemethods generation (optional)
- smaller code for some generated methods (string escaping factorized for instance)
- faster code for some generated methods (less function calls, more constants)
- faster parser (
from*) methods: don't use an intermediate ATerm representation
- pretty-printed generated code, optional compact (no indentation) option
- no hooks
- no ant task
- imports only builtins
- generates no comments
- slightly different command-line arguments syntax