It used to crash for:
Attribute without its name (e.g. Foo<@>)
Missing name with attributes (e.g. Foo<@available(*, unavailable)>
- Synthesize identifier token so that the SyntaxParsingContext can
successfully construct the Attribute syntax.
- Don't throw away the parsed attributes if the parameter name is
missing. Instead, construct GenericParamater syntax anyway, and handle
it in ASTGen.
For the intermediate lookup table of the parsed decl attributes, use
location the first token of the attributes instead of the start location
of the *parsed* attribute list because some attributes can be ignored
during the parsing.
rdar://problem/55952739
Make sure StructureMarkerRAII checks structure nesting level on all paths. Previously swift crashed with no diagnostic on deeply nested '('. Now we print an error when more than 256 parens deep, just as we always have for '['.
fixes SR-4866
This patch also performs minor refactoring to align syntax parsing
context with the right scope. We start to support the generic clauses
because they are necessary pieces to construct struct or
function syntax node.
Because generic where clause doesn't coerce well to our existing syntax
context kinds, we add a new syntax context kind with this patch called
"Syntax". This context kind indicates that when error occurs, the
collection of syntax nodes falling into the context should be coerced
to UnknownSyntax.
We can just use parseType() everywhere instead. We already check
for non-identifier types in inheritance clauses elsewhere, and indeed
we have to anyway because an identifier type might resolve to a
type alias whose underlying type is a non-nominal type.
It doesn't look like this change made any diagnostics worse, but if
we find a case where it did, we could revert it.
We had two slightly different codepaths to diagnose ': class'
in an inheritance clause where it is not supported.
For generic parameters, we would fix the 'class' to 'AnyObject',
but for associated types we didn't do this. Perform the fix in
all cases where it makes sense and remove one of the two
diagnostics.
Complete generic parameters and their members inside generic where
clauses on structs, classes, enums, extensions, typealiases, funcs,
subscripts and inits.
Still not handled correctly are associatedtypes.
rdar://problem/20582394
This doesn't give particularly useful information yet (i.e. Self isn't
listed, see rdar://problem/31981641 ), but it does stop the completion
code from just directly crashing.
Fixes rdar://problem/31981486.
This biggest change is:
- LayoutConstraintInfo is now a FoldingSetNode, which allows for proper canonicalization of LayoutConstraints. This is important for the correctness of type comparisons if types contain layout constraints.
No functionality changes from the client's point of view.
Test for the bogus depth when canonicalizing a reference to the
generic parameters, so we don't canonicalize before the depth has been
set. This flushed out various issues where we weren't validating the
context eagerly enough.
We used to drop the entire generic parameter list if one of the
entries failed to parse. This caused a problem where the generic
parameters were still available for name lookup, so they had
to be special-cased since there's no generic environment set up
in this case.
Now, keep the parts of the generic parameter list around that
parsed successfully.
When I first made the change, almost a hundred crashers regressed;
now all the underlying issues have been fixed.
The result is that in addition to removing a crappy hack we get
some more mileage out of the compiler_crashers, because stuff like
this now builds a generic environment:
class S<T{...}
The new `@_specialize` attribute has a syntax like this:
```swift
@_specialize(exported: true, kind: full, where K == Int, V == Int)
@_specialize(exported: false, kind: partial, where K: _Trivial64)
func dictFunction<K, V>(dict: Dictionary<K, V>) {
}
```
If `exported` is set, the corresponding specialization would have a public visibility and can be used by clients.
If `exported` is omitted, it's value is assumed to be `false`.
If `kind` is `full` it means a full specialization.
If `kind` is `partial` it means a partial specialization.
If `kind` is omitted, its value is assumed to be `full`.
Changes:
* Terminate all namespaces with the correct closing comment.
* Make sure argument names in comments match the corresponding parameter name.
* Remove redundant get() calls on smart pointers.
* Prefer using "override" or "final" instead of "virtual". Remove "virtual" where appropriate.
These APIs return SourceLocs, and eventually the Parser should consume
tokens, which now include source trivia such as whitespace and comments,
and package them into a purely syntactic tree. Just a tiny step. NFC.
Store leading a trailing "trivia" around a token, such as whitespace,
comments, doc comments, and escaping backticks. These are syntactically
important for preserving formatting when printing ASTs but don't
semantically affect the program.
Tokens take all trailing trivia up to, but not including, the next
newline. This is important to maintain checks that statements without
semicolon separators start on a new line, among other things.
Trivia are now data attached to the ends of tokens, not tokens
themselves.
Create a new Syntax sublibrary for upcoming immutable, persistent,
thread-safe ASTs, which will contain only the syntactic information
about source structure, as well as for generating new source code, and
structural editing. Proactively move swift::Token into there.
Since this patch is getting a bit large, a token fuzzer which checks
for round-trip equivlence with the workflow:
fuzzer => token stream => file1
=> Lexer => token stream => file 2 => diff(file1, file2)
Will arrive in a subsequent commit.
This patch does not change the grammar.
