Restructure the ELF handling to be completely agnostic to the OS.
Rather than usng the loader to query the section information, use the
linker to construct linker tables and synthetic markers for the
beginning and of the table. Save off the values of these pointers and
pass them along through the constructor to the runtime for registration.
This removes the need for the begin/end objects. Remove the special
construction of the begin/end objects through the special assembly
constructs, preferring to do this in C with a bit of inline assembly to
ensure that the section is always allocated.
Remove the special handling for the various targets, the empty object
file can be linked on all the targets.
The new object file has no requirements on the ordering. It needs to
simply be injected into the link.
Name the replacement file `swiftrt.o` mirroring `crt.o` from libc. Merge
the constructor and the definition into a single object file.
This approach is generally more portable, overall simpler to implement,
and more robust.
Thanks to Orlando Bassotto for help analyzing some of the odd behaviours
when switching over.
This doc-comment's two paragraphs said in the first that the memory must be uninitialized before assignment, while in the second that it cannot be uninitialized.
After changes to stdlib and type-checker related to removing
of the `_Strideable` it's now possible to properly support
iterator types for *ClosedRange without any workarounds.
Resolves: rdar://problem/25584401
Previously we used @autoclosure on the _VariantDictionaryBuffer methods and forwarded closures directly.
However, this relied on an unintentional type-checker inconsistency, tracked by SR-5719.
Using && here causes us to go down a SILGen path that guarantees that self will
be evaluated over the entire && expression instead of just the LHS. This cause
the uniqueness check to always return false at -Onone. At -O, the optimizer is
smart enough to remove this issue.
rdar://33358110
Solution ranking always prefers methods declared on concerete type
vs. in protocol extension, which means in order to to disambiguite
expressions like `(0 ..< 42).contains(7)` we need to have `contains(_:)`
method on both `Range` and `CountableRange` declared explicitly, that
ensures that `CountableRange` is prefered since `Strideable` inherits
from `Comparable` which makes it "more specialized".
* [runtime] Clean up symbols in error machinery.
* [runtime] Clean up symbols in Foundation overlay.
* [runtime] Clean up symbols in collections and hashing.
* [runtime] Remove symbol controls from the Linux definition of swift_allocError.
* [tests] Add more stub functions for tests that link directly to the runtime.
Currently this is implemented as Float80(literal), which gets interpreted as Float80(Double(literal)), unfortunately. Switch to the correct spelling: literal as Float80.
Resolves rdar://problem/35459284
In the absence of conditional conformances, they are now extra types
with different constraints, plus extra overloads to the
stride(from:to:by:) and stride(from:through:by:).
When we pre-scan the components of a key path pattern to determine its runtime type and instance size, we would short-circuit upon seeing an optional-chaining component, since that makes a key path definitely read-only, but the loop also accumulates the size of the instance we're supposed to allocate, so…bad stuff happened. Leave out the short-circuit, fixing SR-6096 | rdar://problem/34889333 .
Extensions on ImplicitlyUnwrappedOptional are not actually accessible
because we force the optional (and thus can only access things on the
type it is wrapping).
Remove these from the stdlib in order to pave the way toward fully
implementing SE-0054.
