Tags: gjtorikian/sorbet
Tags
Update T::Struct docs (sorbet#6502) * Update T::Struct docs * fixes Co-authored-by: Bart de Water <496367+bdewater@users.noreply.github.com>
Skip checking block type in sorbet-runtime when possible (sorbet#6487) * Fix benchmark `time_block` assumes the thing under test happens twice, unless provided an optional argument saying otherwise. * Add integer_param_and_block benchmark * Stop checking `block_type` in sorbet-runtime The Ruby VM does this for us. * With proper `nil` checking * Allow fast path validators to fire when we can skip block type
Avoid excess allocations in `T::Types::Enum#name` (sorbet#6506) This method is showing up alarmingly high in object allocation profiles on Stripe's codebase. May as well try caching the string to see if this helps. These type objects should not be getting mutated after initialization so this should be fine.
Revert "Make runtime type check `attr_accessor` writers (sorbet#6426)" ( sorbet#6505) This reverts commit b1a7f2c.
Revert "update vscode-languageclient with LSP 3.17 support (sorbet#6492… …)" (sorbet#6501) * Revert "update `vscode-languageclient` with LSP 3.17 support (sorbet#6492)" This reverts commit dbe0dca. * bump version to 0.3.18
delete some dead forward definitions from `shared_ptr<core::Type>` da… …ys (sorbet#6499)
delete unused `blockRetrunTemp` (sorbet#6500)
Implement `Array#dig` and `Hash#dig` intrinsic (sorbet#6491) * Implement `Array#dig` and `Hash#dig` intrinsic * Show new autocorrects * Add more tests * Update core/types/calls.cc Co-authored-by: Nathan Froyd <froydnj@gmail.com> * Add tuple test * Update test exp * Update test/testdata/infer/dig.rb Co-authored-by: Nathan Froyd <froydnj@gmail.com> Co-authored-by: Nathan Froyd <froydnj@gmail.com>
Add error for non-`Proc`-typed block argument (sorbet#6486) * Add error for non-`Proc`-typed block argument This doesn't actually work in the Ruby VM, and neither does it work in sorbet-runtime. * fastmoda 'blk: BasicObject' 'blk: T.untyped' rbi * Proc#call -> `T.untyped` This approximates the behavior that used to happen when `BasicObject` used to be used as a `&blk` arg type. * Update existing tests * Add snapshot before implementing autocorrect * Capture behavior change due to autocorrect * sbg && tools/scripts/update_testdata_exp.sh --no-build test/testdata/namer/yield.rb * Suggestions from code review * Add to_proc + blockpass test
Fix crash in getCallArguments (sorbet#6484) Fixes sorbet#6475 This is causing a handful of background crashes on Stripe's codebase every day. Separately, this pattern seems like it should be banned outright--I've opened a ticket to handle that separately: <sorbet#6483>
PreviousNext