John McCall [Fri, 14 Jan 2011 10:35:38 +0000 (10:35 +0000)]
When simplifying a cleanup's entry by merging it into a single predecessor,
replace all uses of the entry with the predecessor. There are no cleanups
relying on this right now, but if we ever want a cleanup with a phi inside
it, this will be important.
Douglas Gregor [Fri, 14 Jan 2011 02:55:32 +0000 (02:55 +0000)]
Start implementing support for substitution into pack expansions that
involve template parameter packs at multiple template levels that
occur within the signatures members of class templates (and partial
specializations thereof). This is a work-in-progress that is deficient
in several ways, notably:
- It only works for template type parameter packs, but we need to
also support non-type template parameter packs and template template
parameter packs.
- It doesn't keep track of the lengths of the substituted argument
packs in the expansion, so it can't properly diagnose length
mismatches.
However, this is a concrete step in the right direction.
Ted Kremenek [Thu, 13 Jan 2011 20:58:56 +0000 (20:58 +0000)]
Remove warning in dead stores checker for
dead stores within nested assignments. I have
never seen an actual bug found by this specific
warning, and it can lead to many false positives.
Move name mangling support from CodeGen to AST. In the
process, perform a number of refactorings:
- Move MiscNameMangler member functions to MangleContext
- Remove GlobalDecl dependency from MangleContext
- Make MangleContext abstract and move Itanium/Microsoft functionality
to their own classes/files
- Implement ASTContext::createMangleContext and have CodeGen use it
Douglas Gregor [Thu, 13 Jan 2011 18:47:47 +0000 (18:47 +0000)]
Only apply the parameter pack matching of C++0x [temp.arg.template]p3
when we're actually matching a template template argument to a
template template parameter. Otherwise, use strict matching.
Fixes <rdar://problem/8859985> clang++: variadics and out-of-line definitions.
Zhongxing Xu [Thu, 13 Jan 2011 12:30:12 +0000 (12:30 +0000)]
Support inlining base initializers. We still haven't got it completely right,
since the bindings are purged after they are set up. Need to investigate
RemoveDeadBindings algorithm.
Ted Kremenek [Thu, 13 Jan 2011 06:58:15 +0000 (06:58 +0000)]
Fix a corner case in RegionStore where we assign
a struct value to a symbolic index into array.
RegionStore can't actually reason about this,
so we were getting bogus warnings about loading
uninitialized values from the array. The solution
is invalidate the entire array when we cannot
represent the binding explicitly.
Douglas Gregor [Thu, 13 Jan 2011 00:08:50 +0000 (00:08 +0000)]
Implement C++0x [temp.arg.template]p3, which allows slightly fuzzy
matching of variadic template template parameters to template
arguments. This paragraph was the subject of ISO C++ committee
document N2555: Extending Variadic Template Template Parameters.
Douglas Gregor [Wed, 12 Jan 2011 22:41:29 +0000 (22:41 +0000)]
Add the location of the right parenthesis of a C++ named cast
(static_cast, dynamic_cast, reinterpret_cast, or const_cast) to
improve source-location information. Fixes PR8960.
Douglas Gregor [Wed, 12 Jan 2011 22:04:05 +0000 (22:04 +0000)]
Add some more partial-ordering tests, including one that changes with
the proposed resolution to core isue 692. I'm not certain which way
we'll go on this one.
Douglas Gregor [Wed, 12 Jan 2011 17:07:58 +0000 (17:07 +0000)]
Teach TreeTransform how to transform a pack expansion type into
another pack expansion type. This can happen when rebuilding types in
the current instantiation.
Fixes <rdar://problem/8848837> (Clang crashing on libc++ <functional>).
Jay Foad [Wed, 12 Jan 2011 09:06:06 +0000 (09:06 +0000)]
PR3558: mark "logically const" accessor methods in ASTContext as const,
and mark the fields they use as mutable. This allows us to remove a few
const_casts.
John McCall [Wed, 12 Jan 2011 03:41:02 +0000 (03:41 +0000)]
Fix a latent bug where, after emitting an expression statement, we would
delete the block we began emitting into if it had no predecessors. We never
want to do this, because there are several valid cases during statement
emission where an existing block has no known predecessors but will acquire
some later. The case in my test case doesn't inherently fall into this
category, because we could safely emit the case-range code before the statement
body, but there are examples with labels that can't be fallen into
that would also demonstrate this bug.
Douglas Gregor [Tue, 11 Jan 2011 23:09:57 +0000 (23:09 +0000)]
Add TemplateArgument::CreatePackCopy() to create a new parameter pack
in ASTContext-allocated memory, copying the provided template
arguments. Use this new routine where we can. No functionality change.
Douglas Gregor [Tue, 11 Jan 2011 22:21:24 +0000 (22:21 +0000)]
Implement partial ordering of class template partial specializations
and function templates that contain variadic templates. This involves
three small-ish changes:
(1) When transforming a pack expansion, if the transformed argument
still contains unexpanded parameter packs, build a pack
expansion. This can happen during the substitution that occurs into
class template partial specialiation template arguments during
partial ordering.
(2) When performing template argument deduction where the argument
is a pack expansion, match against the pattern of that pack
expansion.
(3) When performing template argument deduction against a non-pack
parameter, or a non-expansion template argument, deduction fails if
the argument itself is a pack expansion (C++0x
[temp.deduct.type]p22).
Rafael Espindola [Tue, 11 Jan 2011 21:44:37 +0000 (21:44 +0000)]
Set unnamed_addr for type infos that we are confortable marking as hidden. I
think it is safe to mark all type infos with unnamed_addr, but I am not sure.
[analyzer] Introduce ObjCSelfInitChecker, which checks initialization methods to verify that they assign 'self' to the
result of an initialization call (e.g. [super init], or [self initWith..]) before using any instance variable or
returning 'self'.
[analyzer] Add 'bool ignorePrefix' parameter to cocoa::deriveNamingConvention to control whether
the prefix should be ignored.
E.g. if ignorePrefix is true, "_init" and "init" selectors will both be result in InitRule, but if
ignorePrefix is false, only "init" will return InitRule.
Douglas Gregor [Tue, 11 Jan 2011 17:34:58 +0000 (17:34 +0000)]
Implement C++ [temp.func.order]p5 more directly, by passing down the
number of explicit call arguments. This actually fixes an erroneous
test for [temp.deduct.partial]p11, where we were considering
parameters corresponding to arguments beyond those that were
explicitly provided.
Ted Kremenek [Tue, 11 Jan 2011 06:37:47 +0000 (06:37 +0000)]
Rework ExprEngine::processCFGBlockEntrance()
to use a node builder. This paves the way
for Checkers to interpose (via a "visit" method)
at the entrance to blocks.
Douglas Gregor [Tue, 11 Jan 2011 03:23:19 +0000 (03:23 +0000)]
Ensure that the result type of an Objective-C class message send is
complete. However, if it returns a reference type, don't require the
type it refers to to be complete. Fixes <rdar://problem/8807070>.
Douglas Gregor [Tue, 11 Jan 2011 03:14:20 +0000 (03:14 +0000)]
When mapping from a function parameter pack to the set of function
parameters it expanded to, map exactly the number of function
parameters that were expanded rather than just running to the end of
the instantiated parameter list. This finishes the implementation of
the last sentence of C++0x [temp.deduct.call]p1.
Douglas Gregor [Tue, 11 Jan 2011 01:52:23 +0000 (01:52 +0000)]
Implement the last bullet of [temp.deduct.type]p5 and part of the last
sentence of [temp.deduct.call]p1, both of which concern the
non-deducibility of parameter packs not at the end of a
parameter-type-list. The latter isn't fully implemented yet; see the
new FIXME.
Douglas Gregor [Tue, 11 Jan 2011 00:27:36 +0000 (00:27 +0000)]
Add testing for unexpanded parameter packs in all of the C++
expression kinds. This is (indirectly) a test verifying that the
recursive AST visitor is visiting the children of these expression
nodes.
Douglas Gregor [Mon, 10 Jan 2011 20:53:55 +0000 (20:53 +0000)]
Implement more of C++0x [temp.arg.explicit]p9, allowing extension of
pack expansions in template argument lists and function parameter
lists. The implementation of this paragraph should be complete
*except* for cases where we're substituting into one of the unexpanded
packs in a pack expansion; that's a general issue I haven't solved yet.
Douglas Gregor [Mon, 10 Jan 2011 17:53:52 +0000 (17:53 +0000)]
Repent for my copy-and-paste sins, factoring out the code that forms
argument packs from a set of deduced arguments, then checks that those
argument packs match previously-deduced argument packs.
Douglas Gregor [Mon, 10 Jan 2011 17:35:05 +0000 (17:35 +0000)]
Factor out the code to set up template argument deduction for a set of
template argument packs. This also ensures that explicitly-specified
template arguments get properly represented in those cases.
Douglas Gregor [Mon, 10 Jan 2011 07:32:04 +0000 (07:32 +0000)]
Work-in-progress implementation of C++0x [temp.arg.explicit]p9, which
allows an argument pack determines via explicit specification of
function template arguments to be extended by further, deduced
arguments. For example:
template<class ... Types> void f(Types ... values);
void g() {
f<int*, float*>(0, 0, 0); // Types is deduced to the sequence int*, float*, int
}
There are a number of FIXMEs in here that indicate places where we
need to implement + test retained expansions, plus a number of other
places in deduction where we need to correctly cope with the
explicitly-specified arguments when deducing an argument
pack. Furthermore, it appears that the RecursiveASTVisitor needs to be
auditied; it's missing some traversals (especially w.r.t. template
arguments) that cause it not to find unexpanded parameter packs when
it should.
The good news, however, is that the tr1::tuple implementation now
works fully, and the tr1::bind example (both from N2080) is actually
working now.
Sean Hunt [Sat, 8 Jan 2011 20:30:50 +0000 (20:30 +0000)]
Renamed CXXBaseOrMemberInitializer to CXXCtorInitializer. This is both shorter,
more accurate, and makes it make sense for it to hold a delegating constructor
call.
projects / clang / log