Previously this would mean something like
class C {
private func f() {}
}
would end up with the symbol for f being completely public (external,
default visibility), even though it only needs to match the internal
class (external, hidden visibility).
Enums with the ns_error_domain attribute represent codes for NSError,
which means Swift developers will expect to interact with them in
terms of Error. SE-0112 improved bridging for these enums to generate
a struct with the following form:
struct MyError: Error {
@objc enum Code: RawRepresentable {
case outOfMemory
case fileNotFound
}
var userInfo: [NSObject: AnyObject] { get }
static var outOfMemory: Code { get }
static var fileNotFound: Code { get }
}
where MyError.Code corresponds to the original MyError enum defined in
Objective-C. Until recently, both the enum and the synthesized struct
were marked as having the original enum as their "Clang node", but
that leads to problems: the struct isn't really ObjC-compatible, and
the two decls have the same USR. (The latter had already been worked
around.)
This commit changes the struct to be merely considered a synthesized
"external definition", with no associated Clang node. This meant
auditing everywhere that's looking for a Clang node and seeing which
ones applied to external definitions in general.
There is one regression in quality here: the generated struct is no
longer printed as part of the Swift interface for a header file, since
it's not actually a decl with a corresponding Clang node. The previous
change to AST printing mitigates this a little by at least indicating
that the enum has become a nested "Code" type.
External private definitions coming from a -sil-serialize-all
module must be effectively public at the LLVM level too, fixing
a linker error when referencing closures from a deserialized
function body.
Also, add a third [serializable] state for functions whose bodies we
*can* serialize, but only do so if they're referenced from another
serialized function.
This will be used for bodies synthesized for imported definitions,
such as init(rawValue:), etc, and various thunks, but for now this
change is NFC.
This gives big code size wins for unused types and also for types, which are never used in a generic context.
Also it reduces the amount of symbols in the symbol table.
The size wins heavily depend on the project. I have seen binary size reductions from 0 to 20% on real world projects.
rdar://problem/30119960
This is NFC in intent, but I had to restructure the code to emit more
of the lists "inline", which means I inevitably altered some IRGen
emission patterns in ways that are visible to tests:
- GenClass emits property/ivar/whatever descriptors in a somewhat
different order.
- An ext method type list is now emitted as just an array, not a struct
containing only that array.
- Protocol descriptors are no longer emitted as packed structs.
I was sorely tempted to stop using packed structs for all the metadata
emission, but didn't really want to update that many tests in one go.
to correctly handle generalized protocol requirements.
The major missing pieces here are that the conformance search
algorithms in both the AST (type substitution) and IRGen
(witness table reference emission) need to be rewritten to
back-track requirement sources, and the AST needs to actually
represent this stuff in NormalProtocolConformances instead
of just doing ???.
The new generality isn't tested yet; I'm looking into that,
but I wanted to get the abstractions in place first.
