Finalize feature specification of strong mode

[anders-sandholm]

My apologies if this issue already exists somewhere. If so, pls merge/close.

Would be great if we could use this as a meta issue to track what is still missing in order for us to have a complete feature specification for strong mode.

This will be useful for our teams implementing strong mode as well as for completing/expanding our co19 test suite.

@floitschG @leafpetersen @eernstg @lrhn @munificent

• New type rules and type system in the formal spec (e.g., sub-typing, ...) (eernst) informal spec by @leafpetersen underway
• Capture "Ambiguities must be explicitly resolved" in a feature spec regarding sub-classing and class hierarchies (eernst) (informal spec underway)
• Feature specification of type inference #?????? (eernst)
• Optional new and const (including auto-const)

@eernstg edited the bullet list above to indicate that auto-const is now included rather than being discussed; the CL has been landed.

Note the following issues whose requests are subsumed by or at least overlapping the requests in this issue:

• Finalize and specify rules for local variable inference #27500 requests local inference specification.
• Specify strong mode "top level" inference #27499 requests global inference specification.
• Interface inheritance and override rules are unclear #16443 closely related to "Ambiguities must be explicitly resolved" (CL on that).
• Language specification: Support generalized void #30597 requests specification of generalized void; a related CL: void-to-void data flow.

[eernstg]

I wrote a CL for extending dartLangSpec.tex with support for generic functions. An obvious next step would be to update all the subtype rules (which includes deleting all references to '<<'/'more specific than').

Apart from that, the most recent additions to the 'informal' directory were created in order to clarify various Dart 2 features, so they would serve as a source of additional topics to handle.

[eernstg]

Work is ongoing: Said CL about generic functions has been landed as 673d5f0. Feature spec about Dart 2 subtyping written by @leafpetersen (addressing the 'New type rules and..' item) is in this CL.

[eernstg]

Added milestone for reaching the level where feature specs have stabilized, but the formal spec may not yet have been updated.

[eernstg]

Wrote a proposed feature spec in this CL for the 'Ambiguities must be explicitly resolved' item.

[eernstg]

This CL addresses the 'Optional new and const' item, may probably land as-is.

Edit: Said CL was landed as 6e3a274.

[eernstg]

Updated the initial message by adding links to CLs aimed at closing bullet points in this issue.

[a-siva]

Somebody needs to own this issue, the assignee field is blank.

[eernstg]

I've assigned myself at this point, but note that it's actually a 'Type: meta' issue (where there is often no assignee) and remaining work is: Land the subtyping feature spec (by @leafpetersen), clarify the situation on inference feature specs (@leafpetersen has written several documents in the area), review the feature spec proposal on conflict resolution (by @eernstg), and decide on the relationship between void and data flow (that is, choosing at most one of the above-mentioned CLs 3a, 3b).

Edit (Feb 20 2018): The language team decided to go with CL 3b above, and abandon 3a.

[dgrove]

Moving to the "Dart 2 Stable" milestone

[eernstg]

Closing: All specification tasks requested here are complete except for inference, and that is handled in the individual issues mentioned above.