Optional properties now get synthesized encodeIfPresent(...) and
decodeIfPresent(...) calls (to avoid encoding needless information, and
to be more accepting on input).
The removed check was dependent on the malloc implementation of the target operating system, because it asks for a real size of the allocated memory block. Usually, there is only a minor over-allocation due to the way how most malloc implementations work. But this is not guaranteed to hold for any malloc implementation.
rdar://problem/32315336
In `FailureDiagnosis::visitSubscriptExpr` if there is only a single
candidate available, verify that the problem is actually related
to the contextual mismatch by type-checking whole subscript without
contextual info, if that returns a type - it's contextual, otherwise
diagnose as incorrect argument type problem.
Resolves: rdar://problem/31977679
This overload allows `String.filter` to return a `String`, and not
`[Character]`.
In the other hand, introduction of this overload makes `[123].filter`
somewhat ambiguous in a sence, that the compiler will now prefer an
implementatin from a more concrete protocol, which is less efficient for
arrays, therefore extra work is needed to make sure Array types fallback
to the `Sequence.filter`.
Implements: <rdar://problem/32209927>
Introduces a _ThreadLocalStorage struct to hold thread-local, but
global resources. Set it up to host a UBreakIterator and a cache key
for resetting text.
UBreakIterators are extremely expensive to create on the fly, so we
store one for each thread. UBreakIterators are also expensive to bind
to new text, so we cache the text it's currently bound to in order to
help avoid it.
The struct can be expanded with more functionality in the future, but
the standard library should only ever use a single key, and thus
everything should go on this struct. The _ThreadLocalStorage struct is
not meant to be copyable, creatable (by anyone else except the
once-per-thread initialize routine), and should accessed through the
pointers it provides.
Future immediate directions could include cashing multiple
UBreakIterators (e.g. avoid a text reset for mutual character
iteration patterns, etc).
Test added in test/stdlib/ThreadLocalStorage.swift.
This is a bit more robust and user-friendly than hoping more brittle recovery in SILGen or IRGen for unsupported components kicks in. rdar://problem/32200714
It is still possible to get the same behavior by providing an explicit
type context, or in a generic code, but otherwise, `String("")!` will
not compile.
Currently, AffineTransform.rotate(byRadians:) simply assigns the sines and cosines to
the transformation matrix, throwing away information about the current scale, rotation, etc.
This patch performs a proper rotation by concatenating the rotation matrix.
codingPath more often than not actually needs to be copied, not just
referenced. This makes a big difference for nested containers and
subobjects, which were getting the wrong codingPath values when asking
for them.
This also adds unit tests for JSONEncoder and PropertyListEncoder to
confirm expected behavior.
A property imported from Objective-C, or marked in Swift with the `dynamic` keyword, doesn't have a vtable slot, so can't be identified that way. Use the ObjC selector as the unique identifier to ascribe equality to such components. Fixes rdar://problem/31768669. (While we're here, throw some more execution tests and a changelog note in.)
* [stdlib] Backward compatibility fix for a flatMap on [String]
Since String started to conform to Collection, the flatMap with a
sequence returning closure is now a better match that the one that
relies on the optional promotion in this code:
[""].flatMap { $0 }
which results in the default type of this expression changing from
[String] to [Character].
Restoring the old behavior in Swift 3 mode by adding a very explicit
overload.
Fixes: <rdar://problem/32024978>
* [stdlib] Fixing another compatibility issue with [String].flatMap
