Fixes a crash when property behavior uses are run through the specializer, since they set up associated type conformances that are not recoverable by name lookup.
from the witness tables for their associations rather than passing
them separately.
This drastically reduces the number of physical arguments required
to invoke a generic function with a complex protocol hierarchy. It's
also an important step towards allowing recursive protocol
constraints. However, it may cause some performance problems in
generic code that we'll have to figure out ways to remediate.
There are still a few places in IRGen that rely on recursive eager
expansion of associated types and protocol witnesses. For example,
passing generic arguments requires us to map from a dependent type
back to an index into the all-dependent-types list in order to
find the right Substitution; that's something we'll need to fix
more generally. Specific to IRGen, there are still a few abstractions
like NecessaryBindings that use recursive expansion and are therefore
probably extremely expensive under this patch; I intend to fix those
up in follow-ups to the greatest extent possible.
There are also still a few things that could be made lazier about
type fulfillment; for example, we eagerly project the dynamic type
metadata of class parameters rather than waiting for the first place
we actually need to do so. We should be able to be lazier about
that, at least when the parameter is @guaranteed.
Technical notes follow. Most of the basic infrastructure I set up
for this over the last few months stood up, although there were
some unanticipated complexities:
The first is that the all-dependent-types list still does not
reliably contain all the dependent types in the minimized signature,
even with my last patch, because the primary type parameters aren't
necessarily representatives. It is, unfortunately, important to
give the witness marker to the primary type parameter because
otherwise substitution won't be able to replace that parameter at all.
There are better representations for all of that, but it's not
something I wanted to condition this patch on; therefore, we have to
do a significantly more expensive check in order to figure out a
dependent type's index in the all-dependent-types list.
The second is that the ability to add requirements to associated
types in protocol refinements means that we have to find the *right*
associatedtype declaration in order to find the associated witness
table. There seems to be relatively poor AST support for this
operation; maybe I just missed it.
The third complexity (so far) is that the association between an
archetype and its parent isn't particularly more important than
any other association it has. We need to be able to recover
witness tables linked with *all* of the associations that lead
to an archetype. This is, again, not particularly well-supported
by the AST, and we may run into problems here when we eliminate
recursive associated type expansion in signatures.
Finally, it's a known fault that this potentially leaves debug
info in a bit of a mess, since we won't have any informaton for
a type parameter unless we actually needed it somewhere.
attached to the representative dependent type in a generic
signature.
This simplifies certain kinds of analysis associated with
generic signatures, especially for minimized generic signatures,
since it means that ArchetypeBuilder::enumerateRequirements
(and hence getCanonicalManglingSignature) will generally use
the same dependent types as the original signature.
Fix some interface type/context type confusion in the AST synthesis from the previous patch, add a unique private mangling for behavior protocol conformances, and set up SILGen to emit the conformances when property declarations with behaviors are visited. Disable synthesis of the struct memberwise initializer if any instance properties use behaviors; codegen will need to be redesigned here.
Parse 'var [behavior] x: T', and when we see it, try to instantiate the property's
implementation in terms of the given behavior. To start out, behaviors are modeled
as protocols. If the protocol follows this pattern:
```
protocol behavior {
associatedtype Value
}
extension behavior {
var value: Value { ... }
}
```
then the property is instantiated by forming a conformance to `behavior` where
`Self` is bound to the enclosing type and `Value` is bound to the property's
declared type, and invoking the accessors of the `value` implementation:
```
struct Foo {
var [behavior] foo: Int
}
/* behaves like */
extension Foo: private behavior {
@implements(behavior.Value)
private typealias `[behavior].Value` = Int
var foo: Int {
get { return value }
set { value = newValue }
}
}
```
If the protocol requires a `storage` member, and provides an `initStorage` method
to provide an initial value to the storage:
```
protocol storageBehavior {
associatedtype Value
var storage: Something<Value> { ... }
}
extension storageBehavior {
var value: Value { ... }
static func initStorage() -> Something<Value> { ... }
}
```
then a stored property of the appropriate type is instantiated to witness the
requirement, using `initStorage` to initialize:
```
struct Foo {
var [storageBehavior] foo: Int
}
/* behaves like */
extension Foo: private storageBehavior {
@implements(storageBehavior.Value)
private typealias `[storageBehavior].Value` = Int
@implements(storageBehavior.storage)
private var `[storageBehavior].storage`: Something<Int> = initStorage()
var foo: Int {
get { return value }
set { value = newValue }
}
}
```
In either case, the `value` and `storage` properties should support any combination
of get-only/settable and mutating/nonmutating modifiers. The instantiated property
follows the settability and mutating-ness of the `value` implementation. The
protocol can also impose requirements on the `Self` and `Value` types.
Bells and whistles such as initializer expressions, accessors,
out-of-line initialization, etc. are not implemented. Additionally, behaviors
that instantiate storage are currently only supported on instance properties.
This also hasn't been tested past sema yet; SIL and IRGen will likely expose
additional issues.
Similarly to how we've always handled parameter types, we
now recursively expand tuples in result types and separately
determine a result convention for each result.
The most important code-generation change here is that
indirect results are now returned separately from each
other and from any direct results. It is generally far
better, when receiving an indirect result, to receive it
as an independent result; the caller is much more likely
to be able to directly receive the result in the address
they want to initialize, rather than having to receive it
in temporary memory and then copy parts of it into the
target.
The most important conceptual change here that clients and
producers of SIL must be aware of is the new distinction
between a SILFunctionType's *parameters* and its *argument
list*. The former is just the formal parameters, derived
purely from the parameter types of the original function;
indirect results are no longer in this list. The latter
includes the indirect result arguments; as always, all
the indirect results strictly precede the parameters.
Apply instructions and entry block arguments follow the
argument list, not the parameter list.
A relatively minor change is that there can now be multiple
direct results, each with its own result convention.
This is a minor change because I've chosen to leave
return instructions as taking a single operand and
apply instructions as producing a single result; when
the type describes multiple results, they are implicitly
bound up in a tuple. It might make sense to split these
up and allow e.g. return instructions to take a list
of operands; however, it's not clear what to do on the
caller side, and this would be a major change that can
be separated out from this already over-large patch.
Unsurprisingly, the most invasive changes here are in
SILGen; this requires substantial reworking of both call
emission and reabstraction. It also proved important
to switch several SILGen operations over to work with
RValue instead of ManagedValue, since otherwise they
would be forced to spuriously "implode" buffers.
When a dependent type is mapped into context, the result will either be
an archetype or a concrete type. The latter occurs if a same-type
constraint exists between the dependent type and the concrete type.
The logic to decide if a type should be passed directly or indirectly
was not handling this case if an interface type was passed down -- we
would just check if there was a class constraint present.
This resulted in mismatching conventions between an interface type and
its corresponding contextual type, which would trigger assertions.
Note that same-type constraints between generic parameters and concrete
types are still not supported for other reasons; the subject of the
constraint must still be an associated type of a generic parameter.
Fixes <rdar://problem/24687460>.
Previously generic parameters constrained to a class were not handled correctly
when accessing associated types in function bodies resulting in strange error
messages mentioning '<<error type>>'.
We already know to print one before printing the body brace (if we're
printing function bodies), and it certainly doesn't belong in the
*name* portion of the decl.
Also add a cursor info test with a deinit.
This splits the printDeclNamEndLoc callback into NameEndLoc and
NameOrSignatureEndLoc variants to differentiate whether or not
signatures are included. All existing clients move to
NameOrSignatureEndLoc to maintain the current behaviour. I'm still not
completely happy with how these are named, but I dont' have any better
ideas right now.
rdar://problem/24292226
...because "build configuration" is already the name of an Xcode feature.
- '#if' et al are "conditional compilation directives".
- The condition is a "conditional compilation expression", or just
"condition" if it's obvious.
- The predicates are "platform conditions" (including 'swift(>=...)')
- The options set with -D are "custom conditional compilation flags".
(Thanks, Kevin!)
I left "IfConfigDecl" as is, as well as SourceKit's various "BuildConfig"
settings because some of them are part of the SourceKit request format.
We can change these in follow-up commits, or not.
rdar://problem/19812930
There was a diagnostic to catch these, but it wasn't triggered
reliably, and it sounds like users were already relying on this
feature working in the few cases where it did.
So instead, just map an archetype's superclass into context
when building the archetype.
Recursion is still not allowed and is diagnosed, for example
<T, U where T : C<U>, U : C<T>>.
Note that compiler_crashers_fixed/00022-no-stacktrace.swift no
longer produces a diagnostic in Sema, despite the fact that the
code is invalid. It does diagnose in IRGen when we map the
type into context. Diagnosing in Sema requires fixing the
declaration checker to correctly handle recursion through a
generic signature. Right now, if recursion is detected, we bail
out, but do not always diagnose. Alternatively, we could
prohibit unbound generic types from appearing in generic
signatures.
This is a more principled fix for rdar://problem/24590570.
There's a group of methods in `DeclContext` with names that start with *is*,
such as `isClassOrClassExtensionContext()`. These names suggests a boolean
return value, while the methods actually return a type declaration. This
patch replaces the *is* prefix with *getAs* to better reflect their interface.
The two types are nearly identical, and Fixnum is only in the Swift branches of LLVM,
not in mainline LLVM.
I do want to add ++ to PointerEmbeddedInt and fix some of this ugliness, but that'll
have to go through LLVM review, so it might take a bit.
Replace pointer arithmetic between multiple base classes with pointer arithmetic
between adjacent allocations. This is still pretty fragile, but at least not
ABI-dependent, and in practice we should generate exactly the same code.
This class formalizes the common case of the "trailing allocation" idiom we use
frequently. I didn't spot any true bugs while making this change, but I did see
places where we were using the wrong pointer type or casting through void* for
no good reason. This will keep us honest.
I'll get to the other libraries soon.
The patch:
1. Removes swift's cpack configuration. We do not use cpack anymore, but LLVM
does so including it twice causes problems. Its simpler to just remove it.
2. Ensures that when building swiftAST standalone, we add
dependencies on Clang's tablegen targets.
The hash function was comparing more fields than operator==, breaking
DenseSets of FoundDeclTy and causing assertion failures.
Incidentally remove the now unused operator< which was only there for
the std::set embedded in the old default implementation of SetVector.
