Golden tests¶

Squiller has a suite of golden tests: inputs with known-good outputs. At the time of writing they are used to test error reporting for incorrect inputs, but they should be extended to testing generated code. Golden tests are a good fit for testing the parser and error reporting, because textual input is easier to construct than manually constructing an AST in a unit test.

Golden tests are located in the golden directory, and can be executed with golden/run.py. See also run.py --help for usage.

Test cases are files with a .test extension. The file consists of the test input, which is fed to sqiller on stdin, then two blank lines, and then the expected output. When the actual output does not match the expected output, run.py prints a diff.