The implementation is making a strong assumption about the array
element type (single-reference layout), but it is extremely
non-obvious that this helper is only called for those element types.
Swift SVN r30183
Leave the qualification off of enum cases and type names when 'print'-ing them, but keep them on 'debugPrint'. (At least, at the outermost level; since ad-hoc printing of structs and tuples uses debugPrint, we'll still get qualification at depth, which kind of sucks but needs more invasive state management in print to make possible.) Implements rdar://problem/21788604.
Swift SVN r30166
This reverts commit r30148, since it regresses the following tests:
Swift :: ClangModules/script.swift
Swift :: Interpreter/currying_protocols.swift
Swift :: Interpreter/if_expr.swift
Swift :: Interpreter/protocol_extensions.swift
Swift :: Interpreter/weak.swift
Swift :: Interpreter/weak_objc.swift
Swift :: PlaygroundTransform/array_did_set.swift
Swift :: PlaygroundTransform/print.swift
Swift SVN r30159
Otherwise, the length of the sequence is in principle lost. If you know
you have a sequence of less than 100 elements, you still want to know
exactly how many elements you initialized from it.
Swift SVN r30104
This is a travesty that makes it hard to reason about tuple vs. scalar
types. It's only use in our own code involved dealing with the
difference in element types between Set and Dictionary in
gyb-generated code. Ewww. Fixes rdar://problem/17963034.
Swift SVN r30074
Like ObjCBool is a legitimate boolean type rather than a typealias for Int8,
DarwinBoolean is better than a typealias for UInt8. It's a BooleanType
(meaning you can use it directly in if/while/?:) and BooleanLiteralConvertible
(meaning you can use 'true' and 'false').
The next commit goes even further, so that you only have to deal with
DarwinBoolean when ABI is important. At all other times it should be
bridged with Bool, just like ObjCBool.
rdar://problem/19013551
Swift SVN r30050
The specialization uses += operator for collections, which performs much
better than the += operator for sequences.
The extend benchmark from rdar://problem/21689050 improves by:
- 2500x with -O
- 15x with -O -whole-module-optimization
Unfortunately we still seem to be about 100x slower than the user's
memcpy version of the same benchmark with -O -whole-module-optimization.
Swift SVN r30033
This brings the David Owens benchmark from http://owensd.io/2015/06/27/performance-xcode7-beta-2.html from parity with simd.h-based C to 3x faster.
Before:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 7.035851 │ 6.304739 │ 9.815832 │ 1.212 │
After:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 2.318357 │ 2.223325 │ 2.697981 │ 0.1490 │
This also addresses rdar://problem/21574425, since Builtin.add_VecNxIntM isn't overflow-checked, and overflow checks really aren't wanted when working with vector types directly.
Reapplying now that Nadav's fixed the ARM64 SelectionDAG issue this exposed before, and Arnold's fixed
yet another SelectionDAG issue exposed after that.
Swift SVN r30006
Metatypes can't directly conform to _ObjectiveCBridgeable, but we can pretend they do by making a struct with the same ABI conform and returning that conformance when we call findBridgeWitness on a metatype. Fixes rdar://problem/16238475.
Swift SVN r29999
This is a straight-up "oops". You could always get to these typealiases via
the protocol, but like the member requirements you should have to say so.
Swift SVN r29952
Revert "simd overlay: Use LLVM vector types."
This reverts commit r29922 and r29924.
More arm64 instruction selection errors.
rdar://21703486
Swift SVN r29941
Remove these standard library types in favor of (T) -> () closures.
It was originally believed that generic optimizations would make these
types profitable, however:
// FIXME: Insert benchmarks here.
rdar://problem/21663799
Swift SVN r29927
This brings the David Owens benchmark from http://owensd.io/2015/06/27/performance-xcode7-beta-2.html from parity with simd.h-based C to 3x faster.
Before:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 7.035851 │ 6.304739 │ 9.815832 │ 1.212 │
After:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 2.318357 │ 2.223325 │ 2.697981 │ 0.1490 │
This also addresses rdar://problem/21574425, since Builtin.add_VecNxIntM isn't overflow-checked, and overflow checks really aren't wanted when working with vector types directly.
Reapplying now that Nadav's fixed the ARM64 SelectionDAG issue this exposed before.
Swift SVN r29922
- Clean up tests
- Create API-specific test structures with clearer labels
- Use `OpaqueValue`s in test collections
- Make tests non-generic to allow array literal declarations
- Remove some unused functions
- Test return values for mutating functions that return a value
- Add expect* test source locations
- Add _customRemoveLast ad-hoc default implementation for `removeLast`
- Add default implementation for reserveCapacity that does nothing.
Swift SVN r29905
This brings the David Owens benchmark from http://owensd.io/2015/06/27/performance-xcode7-beta-2.html from parity with simd.h-based C to 3x faster.
Before:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 7.035851 │ 6.304739 │ 9.815832 │ 1.212 │
After:
RenderGradient ([UInt32].withUnsafeMutablePointer (SIMD)) │ 2.318357 │ 2.223325 │ 2.697981 │ 0.1490 │
This also addresses rdar://problem/21574425, since Builtin.add_VecNxIntM isn't overflow-checked, and overflow checks really aren't wanted when working with vector types directly.
Swift SVN r29891
It was neither testable (because it was internal) nor usable (because in
practice models would be full of ambiguities with methods provided in
extensions for CollectionType.
Swift SVN r29859
ExtensibleCollectionType's operations can all be represented by the
primitive range replacement operation, so fold it into
RangeReplaceableCollectionType.
In addition, provide default implementations of
RangeReplaceableCollectionType's methods.
- New tests added for combinations of (static, generic) calls and
(default, custom) implementations.
- Mark free Swift functions as unavailable with a message to direct the
developer to the protocol methods.
- Mark ExtensibleCollectionType as available with a message added to
direct the developer to the right protocol.
rdar://problem/18220295
Swift SVN r29857
