SIL differentiability witnesses are a new top-level SIL construct mapping
"original" SIL functions to derivative SIL functions.
SIL differentiability witnesses have the following components:
- "Original" `SILFunction`.
- SIL linkage.
- Differentiability parameter indices (`IndexSubset`).
- Differentiability result indices (`IndexSubset`).
- Derivative `GenericSignature` representing differentiability generic
requirements (optional).
- JVP derivative `SILFunction` (optional).
- VJP derivative `SILFunction` (optional).
- "Is serialized?" bit.
This patch adds the `SILDifferentiabilityWitness` data structure, with
documentation, parsing, and printing.
Resolves TF-911.
Todos:
- TF-1136: upstream `SILDifferentiabilityWitness` serialization.
- TF-1137: upstream `SILDifferentiabilityWitness` verification.
- TF-1138: upstream `SILDifferentiabilityWitness` SILGen from
`@differentiable` and `@derivative` attributes.
- TF-20: robust mangling for `SILDifferentiabilityWitness` names.
By convention, most structs and classes in the Swift compiler include a `dump()` method which prints debugging information. This method is meant to be called only from the debugger, but this means they’re often unused and may be eliminated from optimized binaries. On the other hand, some parts of the compiler call `dump()` methods directly despite them being intended as a pure debugging aid. clang supports attributes which can be used to avoid these problems, but they’re used very inconsistently across the compiler.
This commit adds `SWIFT_DEBUG_DUMP` and `SWIFT_DEBUG_DUMPER(<name>(<params>))` macros to declare `dump()` methods with the appropriate set of attributes and adopts this macro throughout the frontend. It does not pervasively adopt this macro in SILGen, SILOptimizer, or IRGen; these components use `dump()` methods in a different way where they’re frequently called from debugging code. Nor does it adopt it in runtime components like swiftRuntime and swiftReflection, because I’m a bit worried about size.
Despite the large number of files and lines affected, this change is NFC.
(implemented by Nathan Hawes @nathawes)
Advance \p Loc to the last non-missing token of the specified or, if it
doesn't contain any, the last non-missing token preceding it in the
tree.
To represent a type with code completion.
type? '.'? <code-completion-token>
This is "parser only" node which is not exposed to SwiftSyntax.
Using this, defer to set the parsed type to code-completion callbacks.
Instead of creating the AST directly in the parser (and libSyntax or
SwiftSyntax via SyntaxParsingContext), make Parser to explicitly create
a tree of ParsedSyntaxNodes. Their OpaqueSyntaxNodes can be either
libSyntax or SwiftSyntax. If AST is needed, it can be generated from the
libSyntax tree.
Instead of creating the AST directly in the parser (and libSyntax or
SwiftSyntax via SyntaxParsingContext), make Parser to explicitly create
a tree of ParsedSyntaxNodes. Their OpaqueSyntaxNodes can be either
libSyntax or SwiftSyntax. If AST is needed, it can be generated from the
libSyntax tree.
ParsedSyntaxBuilder has a convenient function to add member to a syntax-collection
child. The function name uses the type name of the collection's members,
which can lead to name collision. This patch renames it.
This avoids us having to pattern match every source file which should
help speed up the CMake generation. A secondary optimization is
possible with CMake 3.14 which has the ability to remove the last
extension component without having to resort to regular expressions. It
also helps easily identify the GYB'ed sources.
To ensure SwiftSyntax calls a compatible parser library, this patch sets
up a C API that returns a constant string calculated during compilation time to indicate
the version of syntax node declarations. The same hash will be calculated
in the SwiftSyntax (client) side as well by using the same algorithm.
During runtime, SwiftSyntax will verify its hash value is identical to the
result of calling swiftparse_node_declaration_hash before actual
parsing happens.
This patch only sets the API up. The actual implementation of the
hashing algorithm will come later.
Instead of creating syntax nodes directly, modify the parser to invoke an abstract interface 'SyntaxParseActions' while it is parsing the source code.
This decouples the act of parsing from the act of forming a syntax tree representation.
'SyntaxTreeCreator' is an implementation of SyntaxParseActions that handles the logic of creating a syntax tree.
To enforce the layering separation of parsing and syntax tree creation, a static library swiftSyntaxParse is introduced to compose the two.
This decoupling is important for introducing a syntax parser library for SwiftSyntax to directly access parsing.
This reverts commit 121f5b64be.
Sorry to revert this again. This commit makes some pretty big changes. After
messing with the merge-conflict created by this internally, I did not feel
comfortable landing this now. I talked with Saleem and he agreed with me that
this was the right thing to do.
The key thing here is that all of the underlying code is exactly the same. I
purposely did not debride anything. This is to ensure that I am not touching too
much and increasing the probability of weird errors from occurring. Thus the
exact same code should be executed... just the routing changed.
