Suggestion Backlog Slog, 10/5/2015

[RyanCavanaugh]

Planned for discussion today

• 1579 nameof operator

• 1611 contextually type getter return statements in object literal methods

• 2103 rest/spread object properties (Stage 2 ES7 proposal), https://github.com/sebmarkbage/ecmascript-rest-spread

• 2491 allow enums as index signature keys

• 3311 relax rules around super call

• 3411 indexing by a const enum should be the same as a literal (in tuples)

• 3841 constructor property of classes should point to the static side of the class

• 3749 allow var x: typeof expr[0] (useful for arrays and tuples)

• 3915 downlevel emit captured lets in for loops

• 4035 publish TypeScript on Bower

• 4089 when an object literal is contextually typed by an unindexed type, don't set allowNewIdentifiers = true in the LS

• 4241 contextually type function parameters even when other parameters have type annotations

• 4260 allow trailing void parameters to be omitted in function calls

• 4868 more type guards (typeof x === 'function' or 'object')

• 5101 when a function's return expression is a recognized type guard (e.g. x instanceof y), automatically infer the function itself to be a type guard

[RyanCavanaugh]

• 1579 nameof operator

  • Some overlap with memberof proposal we have floating around somewhere
  • What if Rename / Find References solved this instead?
  • Punt and see if related approved suggestions address this sufficiently

• 1611 contextually type getter return statements in object literal methods

  • Won't Fix due to complexity in how we fix parameters during overload resolution and lack of feedback

• 2103 rest/spread object properties (Stage 2 ES7 proposal), https://github.com/sebmarkbage/ecmascript-rest-spread

  • May or may not happen in ES7
  • Wait until it gets closer to landed before we commit to any semantics

• 2491 allow enums as index signature keys

  • Meh.

• 3311 relax rules around super call

  • :+1: disallow access to this in lexically-before super calls

• 3411 indexing by a const enum should be the same as a literal (in tuples)

  • :+1: for const enums only

• 3841 constructor property of classes should point to the static side of the class

  • This isn't actually statically true
  • ... but neither are some other things we say
  • Accepting PRs

• 3749 allow var x: typeof expr[0] (useful for arrays and tuples)

  • :+1: allow numeric and string literals only in the brackets

• 3915 downlevel emit captured lets in for loops

  • We error in this case instead of screwing you over
  • Vlad has a proposal / PR
  • Disallow labelled break in this case, but allow standard break
  • Make sure to hoist vars in the body
  • :+1:

• 4035 publish TypeScript on Bower

  • :+1: accepting PRs

• 4089 when an object literal is contextually typed by an unindexed type, don't set allowNewIdentifiers = true in the LS

  • :-1: we screw this up in the wrong direction too often

• 4241 contextually type function parameters even when other parameters have type annotations

  • Breaking change, but in a good way. "Fixing" change under noImplicitAny
  • :+1: Accepting PRs

• 4260 allow trailing void parameters to be omitted in function calls

  • :+1: Accepting PRs

• 4868 more type guards (typeof x === 'function' or 'object')

  • :+1: Accepting PRs. Do we have any other pending type guard work?

• 5101 when a function's return expression is a recognized type guard (e.g. x instanceof y), automatically infer the function itself to be a type guard

  • Too complex, low-yield. Write isFoo separately (and generically?) to get this to work