Douglas Gregor [Sun, 31 Jan 2010 09:12:51 +0000 (09:12 +0000)]
Rework base and member initialization in constructors, with several
(necessarily simultaneous) changes:
- CXXBaseOrMemberInitializer now contains only a single initializer
rather than a set of initialiation arguments + a constructor. The
single initializer covers all aspects of initialization, including
constructor calls as necessary but also cleanup of temporaries
created by the initializer (which we never handled
before!).
- Rework + simplify code generation for CXXBaseOrMemberInitializers,
since we can now just emit the initializer as an initializer.
- Switched base and member initialization over to the new
initialization code (InitializationSequence), so that it
- Improved diagnostics for the new initialization code when
initializing bases and members, to match the diagnostics produced
by the previous (special-purpose) code.
- Simplify the representation of type-checked constructor initializers in
templates; instead of keeping the fully-type-checked AST, which is
rather hard to undo at template instantiation time, throw away the
type-checked AST and store the raw expressions in the AST. This
simplifies instantiation, but loses a little but of information in
the AST.
- When type-checking implicit base or member initializers within a
dependent context, don't add the generated initializers into the
AST, because they'll look like they were explicit.
- Record in CXXConstructExpr when the constructor call is to
initialize a base class, so that CodeGen does not have to infer it
from context. This ensures that we call the right kind of
constructor.
There are also a few "opportunity" fixes here that were needed to not
regress, for example:
- Diagnose default-initialization of a const-qualified class that
does not have a user-declared default constructor. We had this
diagnostic specifically for bases and members, but missed it for
variables. That's fixed now.
- When defining the implicit constructors, destructor, and
copy-assignment operator, set the CurContext to that constructor
when we're defining the body.
Chandler Carruth [Sun, 31 Jan 2010 07:09:11 +0000 (07:09 +0000)]
Handle instantiation of templates with non-type arguments expressed with an
explicit '&' by introducing an address-of operator prior to checking the
argument's type.
Daniel Dunbar [Sun, 31 Jan 2010 00:41:15 +0000 (00:41 +0000)]
cindex/Python: Turn off showing IDs by default, they are really slow to compute
pending a hash function. Also added a --max-depth argument, handy for timing and
limiting the volume of output.
Daniel Dunbar [Sat, 30 Jan 2010 23:31:40 +0000 (23:31 +0000)]
CIndex: Fix ReportSerializedDiagnostics to honor the DiagnosticClient contract
that diagnostics with a source location should occur inside
{Begin,End}SourceFile.
Note that code completion is currently passing in an invalid LangOptions object
due to its implementation, I need to sort this out with Doug.
Daniel Dunbar [Sat, 30 Jan 2010 21:47:16 +0000 (21:47 +0000)]
ASTUnit: Ensure the CompilerInvocation object used in LoadFromCommandLine is
live as long as the ASTUnit. This is useful for clients which want to maintain
pointers to the LangOptions object which ultimately lives in the
CompilerInvocation, although it would be nice to make all of this ownership
stuff more explicit and obvious.
Daniel Dunbar [Sat, 30 Jan 2010 21:47:07 +0000 (21:47 +0000)]
CompilerInstance: Change to not contain the CompilerInvocation object.
This allows clients to install their own CompilerInvocation object, which is
important for clients that may wish to create references to things like
LangOptions whose lifetime will extend past that of the CompilerInstance.
Benjamin Kramer [Sat, 30 Jan 2010 15:01:47 +0000 (15:01 +0000)]
Use StringRef instead of returning a temporary std::string.
This fixes a really nasty bug in Darwin::getDarwinArchName where we were going
StringRef -> temporary std::string -> StringRef (and return the dead StringRef).
The StringRefs from Triple live as long as the Triple itself, that should be
long enough.
Hopefully 2 of 4 MSVC buildbot failures are gone now.
Ted Kremenek [Fri, 29 Jan 2010 23:32:22 +0000 (23:32 +0000)]
Be a little more permissive than C99: allow 'unsigned' to be used for
the field width and precision of a format specifier instead of just
'int'. This matches GCC, and fixes <rdar://problem/6079850>.
Ted Kremenek [Fri, 29 Jan 2010 22:59:32 +0000 (22:59 +0000)]
Per a suggestion from Cristian Draghici, add a method to FormatSpecifier that returns the expected type of the matching data argument. It isn't complete, but should handle several of the important cases.
Ted Kremenek [Fri, 29 Jan 2010 20:55:36 +0000 (20:55 +0000)]
Switch Sema over to using the new implementation of format string
checking. It passes all existing tests, and the diagnostics have been
refined to provide better range information (we now highlight
individual format specifiers) and more precise wording in the
diagnostics.
Ted Kremenek [Fri, 29 Jan 2010 20:29:53 +0000 (20:29 +0000)]
Enhancements to the alternate (WIP) format string checking:
- Add ConversionSpecifier::consumesDataArgument() as a helper method
to determine if a conversion specifier requires a matching argument.
- Add support for glibc-specific '%m' conversion
- Add an extra callback to HandleNull() for locations within the
format specifier that have a null character
Douglas Gregor [Fri, 29 Jan 2010 19:39:15 +0000 (19:39 +0000)]
Fix reference-binding when we have a reference to const volatile type;
previously, we were allowing this to bind to a temporary. Now, we
don't; add test-cases and improve diagnostics.
Douglas Gregor [Fri, 29 Jan 2010 19:14:02 +0000 (19:14 +0000)]
Fix reference binding of const lvalue references to bit-fields, which
requires a temporary. Previously, we were building an initialization
sequence that bound to the bit-field as if it were a real lvalue. Note
that we previously (and still) diagnose binding of non-const
references to bit-fields, as we should.
There's no real way to test that this code is correct, since reference
binding does not *currently* have any representation in the AST. This
fix should make it easier for that to happen, so I've verified this
fix with...
Added InitializationSequence::dump(), to print an initialization
sequence for debugging purposes.
Douglas Gregor [Fri, 29 Jan 2010 17:15:43 +0000 (17:15 +0000)]
When naming a function template via a qualified-id (or any other way
that ADL is suppressed), we need to build an
UnresolvedLookupExpr. Fixes PR6063, which was hitting Boost headers
pretty hard.
Chandler Carruth [Fri, 29 Jan 2010 10:03:42 +0000 (10:03 +0000)]
Bring driver link order in CMake into alignment with the order in the Makefile.
This includes the fix in r94797 to reflect the new dependency of Sema on
Analysis.
Daniel Dunbar [Fri, 29 Jan 2010 03:22:29 +0000 (03:22 +0000)]
ARM/APCS ABI: Fix some problems with bit-fields in structures. After rereading
the ABI spec, this turns out to simplify the code. We still have some annoying
code which mismatches the spec with regard to empty structures.
Ted Kremenek [Fri, 29 Jan 2010 03:16:21 +0000 (03:16 +0000)]
Alternate format string checking: issue warnings for incomplete format specifiers.
In addition, move ParseFormatString() and FormatStringHandler() from
the clang::analyze_printf to the clang namespace. Hopefully this will
resolve some link errors on Linux.
Douglas Gregor [Fri, 29 Jan 2010 00:41:11 +0000 (00:41 +0000)]
When printing diagnostics in c-index-test, also print source ranges
and fix-it information, so we can see everything in one place. Along
the way, fix a few bugs with deserialization and query of diagnostics
in CIndex.
Ted Kremenek [Thu, 28 Jan 2010 23:39:18 +0000 (23:39 +0000)]
Start fleshing out Sema::AlternateCheckPrintfString():
- Add an anonymous class 'CheckPrintfHandler' which will do the
checking of specific format specifiers
- Add checking for using the '@' conversion specifier outside
an ObjC string literal
- Add checking for null characters within the string
Douglas Gregor [Thu, 28 Jan 2010 06:00:51 +0000 (06:00 +0000)]
Introduce serialization and deserialization of diagnostic information
so that CIndex can report diagnostics through the normal mechanisms
even when executing Clang in a separate process. This applies both
when performing code completion and when using ASTs as an intermediary
for clang_createTranslationUnitFromSourceFile().
The serialized format is not perfect at the moment, because it does
not encapsulate macro-instantiation information. Instead, it maps all
source locations back to the instantiation location. However, it does
maintain source-range and fix-it information. To get perfect fidelity
from the serialized format would require serializing a large chunk of
the source manager; at present, it isn't clear if this code will live
long enough for that to matter.
Ted Kremenek [Thu, 28 Jan 2010 01:18:22 +0000 (01:18 +0000)]
Add placeholder function in Sema for new format string checking logic.
This function will use the format string parsing logic in libAnalysis,
and once it is shown to be better than the current implementation it
will replace AlternateCheckPrintfString() entirely.
projects / clang / log