Each test is a shell script. Tests that share files (e.g., `.cabal` files) are grouped under a common sub-directory of [IntegrationTests]. The framework copies the whole group's directory before running each test, which allows tests to reuse files, yet run independently. A group's tests are further divided into `should_run` and `should_fail` directories, based on the expected exit status. For example, the test `IntegrationTests/exec/should_fail/exit_with_failure_without_args.sh` has access to all files under `exec` and is expected to fail.
Each test is a shell script. Tests that share files (e.g., `.cabal` files) are
grouped under a common sub-directory of [IntegrationTests]. The framework
copies the whole group's directory before running each test, which allows tests
to reuse files, yet run independently. A group's tests are further divided into
`should_run` and `should_fail` directories, based on the expected exit status.
For example, the test
`IntegrationTests/exec/should_fail/exit_with_failure_without_args.sh` has access
to all files under `exec` and is expected to fail.
Tests can specify their expected output. For a test named `x.sh`, `x.out` specifies `stdout` and `x.err` specifies `stderr`. Both files are optional. The framework expects an exact match between lines in the file and output, except for lines beginning with "RE:", which are interpreted as regular expressions.
Tests can specify their expected output. For a test named `x.sh`, `x.out`
specifies `stdout` and `x.err` specifies `stderr`. Both files are optional.
The framework expects an exact match between lines in the file and output,
except for lines beginning with "RE:", which are interpreted as regular
expressions.
[IntegrationTests.hs] defines several environment variables:
