Couple months back I asked for advice on lobste.rs about how to write correctness tests for a language's semantic-level tooling (identifier resolution, arity checking, simple type checking, things of this nature). One person strongly recommended writing the assertions about semantic properties in the language itself, in the input test file (for example, defining a "RefersTo" method that your tests search for in the parse tree & evaluate). I went with that approach and it's been working great! So if you are ever also faced with this incredibly niche design problem I definitely recommend doing that.
=> More informations about this toot | View the thread | More toots from ahelwer@discuss.systems
text/geminiThis content has been proxied by September (3851b).