Suggestion Backlog Slog, 11/14/2016

[RyanCavanaugh]

#11779 Optimize emit for destructuring of array
#12071 Allow leading pipe in union type
#9508 Handle expr === true as equivalent to expr for type guards
#5326 Destructured parameter properties
#11839 Breaking change of implicit any {}
#11542 Allow consts to be typeof comparison operands
#11181 Indicate when widening of literals happens
#10065 Un-optionalizing? Scoped assertions?
#8934 switch (true) { in CFA
#10867 Better error reporting in discriminated union

[RyanCavanaugh]

• Unoptimized destructuring compilation #11779 Optimize emit for destructuring of array
  • Community / Accepting PR for only the case of an array literal of sufficient length
• Suggestion: Allowing leading pipe in union type declarations #12071 Allow leading pipe / leading ampersand in union type
  • Only in type alias declarations?
  • Would we enforce that & and | not mix in this situation?
  • Would type a = | b; mean anything?
    • Looks like a prefix unary operator?
  • Why not trailing | as well?
  • Multiple leading / trailng characters?
  • Rule: any type may start with a single leading | or & without effect
  • Community / Accepting PRs
• Type guard handling with boolean compares #9508 Handle expr === true as equivalent to expr for type guards
  • Don't write === true
• Combining destructuring with parameter properties #5326 Destructured parameter properties
  • I don't write classes often... but when I do, I use parameter properties
  • Where do the modifiers go?
    • Inside?
  • Outside?
  • Certainly not both
  • Destructuring syntax is so bad already
  • Is public like let, or is it something else?
  • 'inside' modifiers are confusing; stick to outside
  • Community / Accepting PRs / Daniel might be bored on a plane
• Compiler should allow assignment of new property in empty object #11839 Breaking change of implicit any {} construction
  • For unannotated uninitialized (or initialized with null/undefined/empty arr/obj) variables
    • Keep this an error under NIA
  • Without NIA, keep these as implicit any
  • Bug for 2.1
• Control flow not formed when using consts #11542 Allow consts to be typeof comparison operands
  • 👍
  • Bug for 2.2
• Present somehow to user that literal type is a widening literal type #11181 Indicate when widening of literals happens
  • 👎 without use cases
• Structural Subtyping with Control flow based type analysis #10065 Un-optionalizing? Scoped assertions?
  • Complex to do this automatically
  • Design Limitation
• Control flow analysis: switch (true) case #8934 switch (true) { in CFA
  • 👎
• Discriminated union types better error reporting #10867 Better error reporting in discriminated union
  • This error really sucks
  • Somewhat complex due to how this is all implemented
  • String literals are just another type, how do we know what the right type is?
  • We could report all the elaborations and drown the user in error text
  • We could detect these discriminated unions heuristically
  • Try it in 2.2

Bonus content

• Add support for literal type subtraction  #12215 Subtraction types
• Proposal: Get the type of any expression with typeof #6606 Get the type of a function call
  • But generics...
  • Implementation sketch available
  • What's @JsonFreeman up to these days?