Bob Nystrom
320c804ecf
Add five new Dart benchmark implementations (binary_trees, delta_blue, fib, for, method_call) under test/benchmark/ and register the Dart runner using the fletch runtime in util/benchmark.py. The benchmarks are ported from existing Wren and JavaScript versions, covering tree traversal, constraint solving, recursive Fibonacci, loop performance, and method call overhead.
This contains the automated validation suite for the VM and built-in libraries.
-
benchmark/- Performance tests. These aren't strictly pass/fail, but let us compare performance both against other languages and against previous builds of Wren itself. -
core/- Tests for the built in core library, mainly methods on the core classes. If a bug is inwren_core.corwren_value.c, it will most likely break one of these tests. -
language/- Tests of the language itself, its grammar and runtime semantics. If a bug is inwren_compiler.corwren_vm.c, it will most likely break one of these tests. This includes tests for the syntax for the literal forms of the core classes. -
limit/- Tests for various hardcoded limits. The language doesn't officially specify these limits, but the Wren implementation has them. These tests ensure that limit behavior is well-defined and tested.