1. Regression tests

There are four kinds of test. The compiler must be run with standard output redirected to /dev/null. All these tests use the standard flx_run harness.
rt
Each good test should generate output which, when compared with a known good run, should be much the same. Good tests typically have several things tested.
dt
Bad tests check the compiler issues a diagnostic. Usually, they have just one error.
bt
Nasty tests should crash. These are typically things I'd like to catch with an improved version of the compiler or run time.
pt
Performance tests measure speed. The result is an overall time in seconds, which should get smaller with each compiler improvement.
dn
Tests with non-deterministic results. Output is printed, not comnpared with expected output.
kf
Tests that should compile, but are known to error out or hang during compilation. These should be fixed in some future version of the compiler.
Start python section to spkgs/flx_regression_tests.py[1 /1 ]
     1: #line 36 "./lpsrc/flx_regress.pak"
     2: UNIT_TESTS = glob.glob('test'+os.sep+'rt*.flx')
     3: UNIT_TESTS.sort()
     4: COMPLETION_TESTS = glob.glob('test'+os.sep+'nd*.flx')
     5: COMPLETION_TESTS.sort()
     6: # these are supposed to fail!
     7: BAD_TESTS = glob.glob('test'+os.sep+'bt*.flx')
     8: BAD_TESTS.sort()
     9: 
    10: unit_tests = UNIT_TESTS
    11: completion_tests = COMPLETION_TESTS
    12: failure_tests = BAD_TESTS
    13: pkg_requires = ['flx_compiler','flx_drivers']
    14: iscr_source = ['lpsrc/flx_regress.pak']
    15: 
    16: weaver_directory = 'doc/test/'
    17: 
    18: 
End python section to spkgs/flx_regression_tests.py[1]


1.1. Good tests
1.2. Bad tests
1.3. Bad tests