The CALayer brittleness in <rdar://problem/17014037> is worse than we thought—we can't r/r *at all* before super.init. Go through some contortions to ensure that, when doing direct stored property access in an initializer, we always base off of a +0 value. I tried fixing this in a more general and principled way using SGFContext::AllowPlusZero, but that introduced miscompiles we don't have the luxury of tracking down right now, so hack a more targeted fix that only affects class initializers.
Swift SVN r18635
CALayer and potentially other framework classes implement their own refcounting
schemes that assume [self retainCount] == 1 at initialization time, a guarantee
SILGen didn't attempt to meet until now. Set a flag in SILGenFunction while
doing initializer delegations to indicate that a 'self' reference can consume
the current 'self' binding, and reinitialize rather than reassign 'self' with
the result of the delegation if it was successfully consumed. Fixes
<rdar://problem/17014037>.
Swift SVN r18608
There's a bit of a reshuffle of the ExplicitCastExpr subclasses:
- The existing ConditionalCheckedCastExpr expression node now represents
"as?".
- A new ForcedCheckedCastExpr node represents "as" when it is a
downcast.
- CoerceExpr represents "as" when it is a coercion.
- A new UnresolvedCheckedCastExpr node describes "as" before it has
been type-checked down to ForcedCheckedCastExpr or CoerceExpr. This
wasn't a strictly necessary change, but it helps us detangle what's
going on.
There are a few new diagnostics to help users avoid getting bitten by
as/as? mistakes:
- Custom errors when a forced downcast (as) is used as the operand
of postfix '!' or '?', with Fix-Its to remove the '!' or make the
downcast conditional (with as?), respectively.
- A warning when a forced downcast is injected into an optional,
with a suggestion to use a conditional downcast.
- A new error when the postfix '!' is used for a contextual
downcast, with a Fix-It to replace it with "as T" with the
contextual type T.
Lots of test updates, none of which felt like regressions. The new
tests are in test/expr/cast/optionals.swift.
Addresses <rdar://problem/17000058>
Swift SVN r18556
Array downcast is an explicit cast written "x as U[]", not an implicit
conversion, so make it a subclass of ExplicitCastExpr. The only
effective change here is that we retain the location of the "as" and
the type as written in the AST. No semantic change.
Swift SVN r18391
at the SIL level. Now, the referent type of a WeakStorageType is always
an optional type, instead of always being the underlying reference. This
allows us to represent both optional types. Before, both of these had the
same AST representation of WeakStorageType(T):
weak var x : T?
weak var x : T!
which doesn't work. Now we represent the optional type explicitly in the
AST and at SIL level. This also significantly simplifies a bunch of code
that was ripping off the optional type and resynthesizing it in other places,
and makes SILGen of weak pointers much more straight-forward by eliminating
the need for emitRefToOptional and emitOptionalToRef entirely (see the diffs
in test/SILGen/weak).
Weak pointers still have problems, but this is a big step forward.
Swift SVN r18312
upcasts."
Reinstate "Restrict the array-bridged conversion to non-verbatim
bridging."
Reinstate "[stdlib] Fix T[].bridgeFromObjectiveC"
Reinstate "[stdlib] Fix T[].bridgeFromObjectiveC"
Reinstate "[stdlib] Move _arrayBridgedDownCast to Foundation"
Reinstate "Replace "can" with "cannot" in a message."
Reinstate "Implement support for non-verbatim T[] -> AnyObject[]
upcasts."
This reinstates commit r18291.
This reinstates commit r18290.
This reinstates commit r18288.
This reinstates commit r18287.
This reinstates commit r18286.
This reinstates commit r18293.
This reinstates commit r18283.
John fixed the issue in r18294.
Swift SVN r18299
Revert "Restrict the array-bridged conversion to non-verbatim bridging."
Revert "[stdlib] Fix T[].bridgeFromObjectiveC"
Revert "[stdlib] Fix T[].bridgeFromObjectiveC"
Revert "[stdlib] Move _arrayBridgedDownCast to Foundation"
Revert "Replace "can" with "cannot" in a message."
Revert "Implement support for non-verbatim T[] -> AnyObject[] upcasts."
This reverts commit r18291.
This reverts commit r18290.
This reverts commit r18288.
This reverts commit r18287.
This reverts commit r18286.
This reverts commit r18293.
This reverts commit r18283.
Sorry for the number of reverts, but I needed to do this many to get a clean
revert to r18283.
Swift SVN r18296
One of my recent type-checker changes starting passing all array
upcasts, including "bridged" ones like String[] -> AnyObject[],
through the library entry point _arrayUpCast, which is only meant for
class/ObjC existential types that are bridged verbatim.
Fixes part of <rdar://problem/16952238>; more cleanup to follow.
Swift SVN r18283
If a source file contains the main class for its module, then implicitly emit a top_level_code that invokes UIApplicationMain with the name of the marked class.
Swift SVN r18088
- Continue adding support for checked downcasts of array types (rdar://problem/16535104)
- Fix non-bridged array conversions post-r17868
- Fix rdar://problem/16773693
- Add tests for NSArray coercions to and from Array<T>
Swift SVN r17957
The bridging function takes the original value at +1, so it needs to be consumed so it isn't released in the caller. Fixes <rdar://problem/16877676>.
Swift SVN r17944
Generalize the bridging of AnyObject[] to NSArray into bridging T[] to
NSArray, following what we already have for dictionaries. Use some new
Foundation-level entry points for the bridging, which could use an
audit from someone who understands Array better than I do.
Part of <rdar://problem/16535097>.
Swift SVN r17869
for capture lists and add testcases for stuff that works.
Still more to come, weak pointers need work and one unowned pointer case doesn't work.
Swift SVN r17860
There were two problems here: emitRValueForDecl wasn't doing a semantic
conversion to an RValue when loading from let vardecls of reference storage
type, and we were trying to capture the rvalue type (not the storage type)
when forming the capture.
Swift SVN r17854
'unowned let' initializations don't drop their ReferenceStorageInitialization
transformation. This defines away another class of problems related to weak and
unowned pointers, and allows us to remove a hack added to paper over rdar://16681656.
Swift SVN r17846
Eliminate the duplicate half-broken bridging logic in emitClassConstructorAllocator by referencing foreign initializers through their foreign-to-native thunks, which SILGen knows how to emit already. Do the same thing for factory initializers by suppressing their normal allocating initializer codegen and just referencing the usual foreign-to-native thunk for them. This fixes <rdar://problem/16853205> because we get the ownership thunking right now.
Swift SVN r17840
This keeps us from prematurely releasing values before they're needed by the underlying initializer, which caused a crash if you tried to pass a Swift array into NSArray(objects:count:) via pointer bridging. <rdar://problem/16738050>
Swift SVN r17812
This mostly falls out from the metatype cast infrastructure, but we need to generalize some Sema and SILGen code to accept AnyMetatypeType. Concrete-to-existential metatypes will need more runtime checking that isn't implemented, so raise a 'not implemented' error on those for now.
Swift SVN r17798
This was an artifact of their long-dead C++-based implementations that was way past due to clean up. It's also a convenient workaround for <rdar://problem/16835447>, since we no longer have to look up constructor members for String at SILGen time and potentially lazily instantiate Clang-imported types during SILGen. This totally isn't a real solution to that problem, but it puts out the fire for now.
Swift SVN r17643
Subscript declarations were still encoding the names of index
variables in the subscript type, which unintentionally made them
keyword arguments. Bring subscript declarations into the modern day,
using compound names to encode the subscript argument names, which
provides consistency for the keyword-argument world
<rdar://problem/14462349>. Note that arguments in subscripts default
to not being keyword arguments, which seems like the right default.
We now get keyword arguments for subscripts, so one can overload
subscripts on the names of the indices, and distinguish at the call
site. Under -strict-keyword-arguments, we require strictness here as well.
The IRGen/IDE/SILGen test updates are because the mangling of common
subscripts changed from accidentally having keyword arguments to not
having keyword arguments.
Swift SVN r17393
There were a bunch of things broken here--it's amazing this ever appeared to work.
- Retain 'self' before partial_applying it to the method, so we don't overrelease it.
- Correctly lower the ownership conventions of the dynamic method against the SILDeclRef, so we don't overrelease arguments or over-over-release self, and we handle ObjC methods with weird conventions correctly.
- Thunk when there are bridging type differences between the partially-applied ObjC method and a Swift method, so we don't crash if the method takes NSStrings or other bridged types.
Add verifier checks that the result of 'dynamic_method' and BB arg of 'dynamic_method_br' actually match the method they're dispatching.
Swift SVN r17198
Make ObjC method partial applications go through a native-to-foreign thunk so that we properly handle bridging conversions in the resulting function value. Partial applications through dynamic lookup are still broken because they apparently go through a different path.
Swift SVN r17108
Trying to use a @weak slot as an in-place initialization buffer for an Optional is never going to end well, so refuse to offer reference storage slots as buffers for expr results. Fixes <rdar://problem/16681656>.
Swift SVN r16956
This is part of our poor-man's internal compiler SPI hiding
in the standard library. We don't want these functions showing
up in code completion, etc.
Swift SVN r16916
An unsafe cast from a base to a derived class isn't really all that different from one from Builtin.NativeObject to an arbitrary class, so relax this pair of instructions to allow an arbitrary bitcast. This only combines the instructions; it doesn't attempt to simplify any codegen that was emitting round-trip casts before yet.
Swift SVN r16736
This was part of the original weak design that
there was never any particular reason to rush the
implementation for. It's convenient to do this now
so that we can use it to implement Unmanaged<T> for
importing CF types.
Swift SVN r16693
them with uses of TypeExpr instead. The remaining uses of
MetaTypeExpr (which will be renamed soon) are places where we
are applying the ".dynamicType" virtual property to an expression.
Unadorned uses of types in code, e.g. the Int in "Int.self" are
now represented with TypeExpr.
One unfortunate travesty that doing this work revealed is that we
are extremely sloppy and terrible about maintaining location information
in implicitly generated decls, and our invariants vary quite a bit. This
is really horrible, but I'm not sure whether I'll go fix the hacks or not.
This patch perpetuates the existing crimes, but makes them more visible.
NFC!
Swift SVN r16646
This will represent the return convention of imported __attribute__((objc_returns_inner_pointer)) methods. Leave it unimplemented for now until we can autorelease things sanely.
Swift SVN r16628
when resolving identifiers into types. This will eventually allow us to
solve annoying issues like rdar://15295763&15588967 by better modeling
what we already have.
Swift SVN r16620
