There are a few hairy features (and bugs) that are observable directly or indirectly when running the examples in @roman-spiridorov's coolc project: series of errors in a 'coolc' source file being fed to a 'coolc' parser which doesn't handle all of them in an intuitive error recovery way (as these errors are close together and coding error recovery in a grammar is no sine cure either...)
Plus at least one example in there exhibits a bug which is now fixed (see also #21), but took at least three attempts before my brain caught on and fixed it right -- or so I believe. Anyway: regression of that problem is a high probability and should be regression tested in the jison test set to ensure we're not silently failing in the future.
There are a few hairy features (and bugs) that are observable directly or indirectly when running the examples in @roman-spiridorov's coolc project: series of errors in a 'coolc' source file being fed to a 'coolc' parser which doesn't handle all of them in an intuitive error recovery way (as these errors are close together and coding error recovery in a grammar is no sine cure either...)
Plus at least one example in there exhibits a bug which is now fixed (see also #21), but took at least three attempts before my brain caught on and fixed it right -- or so I believe. Anyway: regression of that problem is a high probability and should be regression tested in the jison test set to ensure we're not silently failing in the future.