- Convert `Protocol.txt` to `Protocol.md`, and use valid Markdown
syntax.
- Spruce up the documentation in the "code completion": more details on
how to test via the REPL, clearer sentences, etc.
- Fix an error in which `sourcekitd-test -req=complete -offset=<offset>`
used the incorrect option `-cc-offset`.
In the new code-completion code path, force any known operators to go
through a fixed sort order. To identify operators unambiguously, add a
new BuiltinOperator code-completion kind to handle non-decl operators
(!, ., ?., and =).
rdar://problem/25994246
rdar://problem/23440367
Fold UnitTest into the subkind, now that it's a bitself, and then remove
the unnecessary inheritance for IndexSymbol and its SourceKit indexing
equivalent.
Eventually we should just have one IndexSymbol, with all the
information. Once we kill FuncDeclIndexSymbol and can get rid of the
enum, we will regain most of the bytes we lost by inlining receiverUSR.
As a first step toward having more detailed "role" information, thread
through a SymbolRole bitset. For now it just contains the existing ref
vs. definition status.
This sort is only for codeComplete, not for codeCompleteOpen, but we
were doing it indiscrimiately leading to ~15% performance penalty in the
latter code path.
rdar://problem/25887631
Having a single interface for source files and modules and having to
pass in a buffer id was a crappy API. This splits the interface to take
either a Module or a SourceFile and handles the buffer id stuff
internally.
The goal is to be able to move the core IndexSwiftASTWalker code out of
SourceKit, leaving only the serialization bits behind.
Mostly this replaces some direct uses of UIdent strings with explicit
enums, and then adds the translation code to produce those enums and to
convert them into UIdents in SourceKit.
rdar://problem/22348041
These results are cached, so we can't use the type-relation. Instead we
use a small hack of checking the textual return type for "Void". This
is obviously not ideal, but it lets us detect the most important cases.
rdar://problem/22810741
As implied in rdar://24818863, striking through a module name may be an overkill to suggest the module is redundant to import. We try to
fine-grain not-recommended-reason so that proper UI cue can be adopted in the future.
Add a dependency on LLVMCore to satisfy the link dependency. Upstream changes
to the Target structure have caused a missing definition:
Undefined symbols for architecture ...:
"llvm::DataLayout::~DataLayout()", referenced from:
clang::TargetInfo::~TargetInfo() in libclangBasic.a(TargetInfo.cpp.o)
Instead of using 'key.usr' and 'key.synthesizedusr', we start to use 'key.usr' and 'key.original-usr' so
that 'key.usr' is consistently being the unique ID for a code entity.
