☂️ Support Wildcards (in generics)

[HosseinYousefi]

• Type "inference" for generics #705
• Better checks for types in generics #704
• Descriptor doesn't represent the actual type edge case #693
• Multilevel deep nested generic classes support #713

[mahesh-hegde]

Related Generics issues:

#705
#704
#693
#713

It might be better to map the JLS generics to Dart generics upfront for once, in form of a specification.

[mahesh-hegde]

I face this problem when implementing parsing of wildcard in asm (TypeUsageSignatureVisitor)¹. So I am looking for a way to solve this if possible.

What are our options here?

1. ? extends QualifiedName will map to QualifiedName and ? super QualifiedName will map to Object. This is probably easiest to implement, and with current implementation of TypeSignatureVisitor in summarizer, this is what happens anyway.

2. ? extends QualifiedName will create a new type variable W1 extends QualifiedName. ? super QualifiedName is still Object.

I need to study the generics spec, especially the covariance part, but I would like to listen from those who already have. :)

Footnotes

1. Map<K, V> uses many ? extends K kind of stuff. Previously +K would get parsed as K but if we want to parse it as ? extends K, we should handle Kind.WILDCARD properly in the generator. ↩