Mike Lischke

As the environment under which the extension runs can be quite different depending on the platform etc. no attempt is made to locate a Java installation. Instead the extension relies...

Hmm, I don't get the idea of the tryXX methods. If getToken would just return a null instead of throwing everything would be fine (in fact I modified the local...

Yes, I got the point about the null check, **but** this is how we do it in every other target. Why does the TS target go a different route (without...

Hmm, yes, I agree. As we see you would use the same return type for `T*` and `T+` which are similar like `T?` and `T`, respectively, which makes me think...

I have to reopen this issue again, because the provided solution doesn't work. In case of syntax errors I have no way of checking if a token exists, because even...

`tryGetRuleContext()` is not a nice replacement for normal rule context retrieval, because it requires the user to lookup the index of the context manually (which is otherwise done on code...

> If your parse operation had syntax errors, don't run the visitor at all. That's not an option. In order to have a passably useful symbol table you have to...

@BurtHarris. I fully support your thoughts here, however how can you return `undefined` when you haven't declared a return value of `T | undefined`? I just tried. bit TS won't...

@sharwell, How's not throwing an exception corrupting the type safety in antlr4ts? You also return `undefined` for optional elements, without having trouble with type safety. I agree you shouldn't optimize...

OK, I see we obviously have different goals and we should stop endlessly arguing. I can just fork antlr4ts and change that to what I need/want, no problem. Thanks.