r194052 - Sema: Improve comment introduced in r193397

Author: majnemer Date: Tue Nov 5 02:01:18 2013 New Revision: 194052 URL: http://llvm.org/viewvc/llvm-project?rev=194052&view=rev Log: Sema: Improve comment introduced in r193397 Modified: cfe/trunk/lib/Sema/SemaLambda.cpp Modified: cfe/trunk/lib/Sema/SemaLambda.cpp URL: http://llvm.org/vie

Re: r193397 - Sema: Do not allow lambda expressions to appear inside of constant expressions

Should be a bit improved in r194052. On Sun, Nov 3, 2013 at 9:30 PM, Richard Smith wrote: > > > > On Fri, Oct 25, 2013 at 2:12 AM, David Majnemer > wrote: > >> Author: majnemer >> Date: Fri Oct 25 04:12:52 2013 >> New Revision: 193397 >> >> URL: http://llvm.org/viewvc/llvm-project?rev=193397&v

Re: [PATCH] [OPENMP] Fix for parsing OpenMP directives with extra braces, brackets and parens

- Original Message - > > > > Comment at: include/clang/Parse/Parser.h:638 > @@ -637,3 +637,3 @@ >/// SkipToTok is specified, it calls SkipUntil(SkipToTok). > Finally, true is >/// returned. >bool ExpectAndConsume(tok::TokenKind ExpectedTok, unsigned Diag,

Re: [PATCH] [OPENMP] Fix for parsing OpenMP directives with extra braces, brackets and parens

Alexey, Can you please post a separate patch for the bit-field SkipUntil change (with no overall functionality change), and then a version of this patch on top of it. They should be two separate commits (although we can do concurrent review). Thanks again, Hal - Original Message - > Fi

r194054 - Enable links now that they work.

Author: rsmith Date: Tue Nov 5 02:27:00 2013 New Revision: 194054 URL: http://llvm.org/viewvc/llvm-project?rev=194054&view=rev Log: Enable links now that they work. Modified: cfe/trunk/www/cxx_status.html Modified: cfe/trunk/www/cxx_status.html URL: http://llvm.org/viewvc/llvm-project/cfe/

Re: Produce direct calls instead of alias to linkonce_odr functions

>> What happens is that at the end of GetOrCreateLLVMFunction we check if >> we have a decl defined in class and if so add it to >> DeferredDeclsToEmit. The net result is that we end up running >> replaceAllUsesWith twice when something in DeferredDeclsToEmit causes >> a new use of the destructor/c

Re: [PATCH] Introduce ClangTool::buildASTs, and buildASTFromCode.

- Address reviewer comments Hi klimek, http://llvm-reviews.chandlerc.com/D2097 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D2097?vs=5336&id=5347#toc Files: include/clang/Tooling/Tooling.h lib/Tooling/Tooling.cpp unittests/Tooling/ToolingTest.cpp Index: include/clang/Too

Re: [PATCH] Introduce ClangTool::buildASTs, and buildASTFromCode.

Comment at: include/clang/Tooling/Tooling.h:79 @@ -63,1 +78,3 @@ + bool runInvocation(clang::CompilerInvocation *Invocation, + FileManager *Files); Manuel Klimek wrote: > Does doxygen default to having the base class' comment here? It does.

r194055 - C++1y sized deallocation: if we have a use, but not a definition, of a sized

Author: rsmith Date: Tue Nov 5 03:12:18 2013 New Revision: 194055 URL: http://llvm.org/viewvc/llvm-project?rev=194055&view=rev Log: C++1y sized deallocation: if we have a use, but not a definition, of a sized deallocation function (and the corresponding unsized deallocation function has been decl

r194056 - Remove stray comma from documentation.

Author: rsmith Date: Tue Nov 5 03:12:55 2013 New Revision: 194056 URL: http://llvm.org/viewvc/llvm-project?rev=194056&view=rev Log: Remove stray comma from documentation. Modified: cfe/trunk/www/cxx_status.html Modified: cfe/trunk/www/cxx_status.html URL: http://llvm.org/viewvc/llvm-projec

Re: r194000 - Use aliases for more constructors and destructors.

Hi, This patch causes Assertion failed: !isAlreadyCodeGenerating && "Error: Recursive compilation detected!", file ..\..\..\..\lib\ExecutionEngine\JIT\JIT.cpp, line 467 and fixed by setting getCodeGenOpts().CXXCtorDtorAliases = 0; Is this the right fix or should something else in the JIT log

r194059 - Lex: Require that '#' be followed by a macro parameter name when preceded by '##'

Author: majnemer Date: Tue Nov 5 03:30:17 2013 New Revision: 194059 URL: http://llvm.org/viewvc/llvm-project?rev=194059&view=rev Log: Lex: Require that '#' be followed by a macro parameter name when preceded by '##' After lexing a '##', we would look ahead and check to see if it was followed by

r194061 - Simplify test case added in r194059

Author: majnemer Date: Tue Nov 5 04:05:58 2013 New Revision: 194061 URL: http://llvm.org/viewvc/llvm-project?rev=194061&view=rev Log: Simplify test case added in r194059 Modified: cfe/trunk/test/Preprocessor/macro_paste_bad.c Modified: cfe/trunk/test/Preprocessor/macro_paste_bad.c URL: htt

r194068 - Do not allow functions or kernels called 'main' in OpenCL.

Author: joey Date: Tue Nov 5 06:30:39 2013 New Revision: 194068 URL: http://llvm.org/viewvc/llvm-project?rev=194068&view=rev Log: Do not allow functions or kernels called 'main' in OpenCL. Modified: cfe/trunk/include/clang/Basic/DiagnosticSemaKinds.td cfe/trunk/lib/Sema/SemaDecl.cpp

r194070 - Fix test that was not being run

Author: alp Date: Tue Nov 5 06:45:40 2013 New Revision: 194070 URL: http://llvm.org/viewvc/llvm-project?rev=194070&view=rev Log: Fix test that was not being run Modified: cfe/trunk/test/Frontend/verify3.c Modified: cfe/trunk/test/Frontend/verify3.c URL: http://llvm.org/viewvc/llvm-project/

r194071 - Make test run reliably

Author: alp Date: Tue Nov 5 06:45:43 2013 New Revision: 194071 URL: http://llvm.org/viewvc/llvm-project?rev=194071&view=rev Log: Make test run reliably llc waits for input on stdin, which was not provided in this test. It was running only thanks to a quirk in the way lit concatenates commands.

r194069 - Fix test that wasn't testing anything

Author: alp Date: Tue Nov 5 06:45:37 2013 New Revision: 194069 URL: http://llvm.org/viewvc/llvm-project?rev=194069&view=rev Log: Fix test that wasn't testing anything Modified: cfe/trunk/test/Analysis/html-diags-multifile.c Modified: cfe/trunk/test/Analysis/html-diags-multifile.c URL: http

r194072 - Fix a test that was never being run, mark it XFAIL

Author: alp Date: Tue Nov 5 06:45:46 2013 New Revision: 194072 URL: http://llvm.org/viewvc/llvm-project?rev=194072&view=rev Log: Fix a test that was never being run, mark it XFAIL Turns out this has always been crashing since the test landed in r186647: Assertion failed: (Access != AS_none &&

r194073 - Disable the test that crashes

Author: alp Date: Tue Nov 5 06:55:00 2013 New Revision: 194073 URL: http://llvm.org/viewvc/llvm-project?rev=194073&view=rev Log: Disable the test that crashes XFAIL wasn't enough for some of the build servers. Modified: cfe/trunk/test/OpenMP/threadprivate_ast_print.cpp Modified: cfe/trunk/

Re: r194072 - Fix a test that was never being run, mark it XFAIL

Hello Alexey, I've been validating the LLVM/clang test suite and noticed that this test was never working. When fixed, it crashes. I've disabled it in SVN, could you investigate? Thanks, Alp. On 05/11/2013 12:45, Alp Toker wrote: > Author: alp > Date: Tue Nov 5 06:45:46 2013 > New Revision: 19

Re: [PATCH] Unknown-bound array static data member of template should be type dependent

A gentle ping. Any inputs on this patch? Does this look good to commit? Thanks! http://llvm-reviews.chandlerc.com/D2049 ___ cfe-commits mailing list cfe-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

r194076 - Disable test fully with REQUIRES

Author: alp Date: Tue Nov 5 07:27:19 2013 New Revision: 194076 URL: http://llvm.org/viewvc/llvm-project?rev=194076&view=rev Log: Disable test fully with REQUIRES Third time's the charm, sorry for the noise. Modified: cfe/trunk/test/OpenMP/threadprivate_ast_print.cpp Modified: cfe/trunk/tes

[clang-tools-extra] r194079 - [extra] pp-trace - Fixed problems found while writing docs.

Author: jtsoftware Date: Tue Nov 5 08:16:11 2013 New Revision: 194079 URL: http://llvm.org/viewvc/llvm-project?rev=194079&view=rev Log: [extra] pp-trace - Fixed problems found while writing docs. Modified: clang-tools-extra/trunk/pp-trace/PPCallbacksTracker.cpp clang-tools-extra/trunk/pp

[libcxx] r194080 - Refactor floating point code for num_get::do_get into a template. No functionality change

Author: marshall Date: Tue Nov 5 08:28:52 2013 New Revision: 194080 URL: http://llvm.org/viewvc/llvm-project?rev=194080&view=rev Log: Refactor floating point code for num_get::do_get into a template. No functionality change Modified: libcxx/trunk/include/locale Modified: libcxx/trunk/inclu

[clang-tools-extra] r194081 - pp-trace user documentation - beginnings

Author: jtsoftware Date: Tue Nov 5 08:41:29 2013 New Revision: 194081 URL: http://llvm.org/viewvc/llvm-project?rev=194081&view=rev Log: pp-trace user documentation - beginnings Modified: clang-tools-extra/trunk/docs/pp-trace.rst Modified: clang-tools-extra/trunk/docs/pp-trace.rst URL: http

Re: r194000 - Use aliases for more constructors and destructors.

Can you provide a test case? Could you also try the patch I emailed about using the aliasee directly when possible? On Tuesday, November 5, 2013, Yaron Keren wrote: > Hi, > > This patch causes > > Assertion failed: !isAlreadyCodeGenerating && "Error: Recursive > compilation detected!", file ..\.

[PATCH] More duplicate code removal in

Hoist duplicate code into two templates: num_get::__do_get_signed () and num_get::__do_get_unsigned () and make the num_get calls call them. No functionality change. -- Marshall Marshall Clow Idio Software A.D. 1517: Martin Luther nails his 95 T

Re: [PATCH] Fix vbtable indices if a class shares the vbptr with a non-virtual base

Addressed the comments Hi rnk, http://llvm-reviews.chandlerc.com/D2089 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D2089?vs=5332&id=5348#toc Files: include/clang/AST/VTableBuilder.h lib/AST/VTableBuilder.cpp lib/CodeGen/CGCXX.cpp lib/CodeGen/CGDebugInfo.cpp lib/CodeGe

r194082 - Fix vbtable indices when a class shares the vbptr with a non-virtual base

Author: timu Date: Tue Nov 5 09:54:58 2013 New Revision: 194082 URL: http://llvm.org/viewvc/llvm-project?rev=194082&view=rev Log: Fix vbtable indices when a class shares the vbptr with a non-virtual base Modified: cfe/trunk/include/clang/AST/VTableBuilder.h cfe/trunk/lib/AST/VTableBu

Re: [PATCH] Fix vbtable indices if a class shares the vbptr with a non-virtual base

Comment at: lib/AST/VTableBuilder.cpp:3314-3316 @@ +3313,5 @@ +const ASTRecordLayout &BaseLayout = Context.getASTRecordLayout(CurBase); +if (!BaseLayout.hasVBPtr() || +DerivedVBPtrOffset != BaseOffset + BaseLayout.getVBPtrOffset()) + break; +

Re: [PATCH] Fix vbtable indices if a class shares the vbptr with a non-virtual base

Landed as r194082. Thanks for the review! http://llvm-reviews.chandlerc.com/D2089 ___ cfe-commits mailing list cfe-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Re: [clang-tools-extra] r194079 - [extra] pp-trace - Fixed problems found while writing docs.

Is there missing test coverage for these cases? On Tue, Nov 5, 2013 at 6:16 AM, John Thompson < john.thompson.jtsoftw...@gmail.com> wrote: > Author: jtsoftware > Date: Tue Nov 5 08:16:11 2013 > New Revision: 194079 > > URL: http://llvm.org/viewvc/llvm-project?rev=194079&view=rev > Log: > [extra

Fix for PR17771

Off-by-one error in the string resizing code. -- Marshall Marshall Clow Idio Software A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait). -- Yu Suzuki PR17771.patch Description: Binary

Re: [PATCH][StaticAnalyzer] new check comparing equal expression

I got a chance to look at the results of analyzing LLVM/Clang with this checker, and also found zero reports—true or false positives. On the one hand, these mistakes probably don't last long in an open-source project. On the other, is it worth having a checker that catches problems that don't show

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

In addition to the performance issues, libc++ apparently can't be found by clang: http://llvm.org/bugs/show_bug.cgi?id=17821 On Sat, Nov 2, 2013 at 10:35 PM, Bob Wilson wrote: > I’m not familiar with those performance issues, so I’m CC’ing Howard in > case he can provide any details on those is

Re: [PATCH][StaticAnalyzer] new check comparing equal expression

On Nov 5, 2013, at 9:26 AM, Jordan Rose wrote: > I got a chance to look at the results of analyzing LLVM/Clang with this > checker, and also found zero reports—true or false positives. On the one > hand, these mistakes probably don't last long in an open-source project. On > the other, is it

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

When I use tot clang, (and want libc++), I will forbid clang from looking the std::headers, and then tell it exactly where the headers are, e.g.: $ clang++ -stdlib=libc++ -std=c++11 -nostdinc++ -I../libcxx/include test.cpp Howard On Nov 5, 2013, at 12:35 PM, Nico Weber wrote: > In addition t

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

On OS X the libc++ headers are meant to be installed along with clang. If you check out the libc++ sources in the LLVM projects directory, they should be installed in the right place. On Nov 5, 2013, at 9:42 AM, Howard Hinnant wrote: > When I use tot clang, (and want libc++), I will forbid cl

Re: Unavailable method checker

On Oct 31, 2013, at 5:43 , Richard wrote: > On 29 Oct 2013, at 17:11, Jordan Rose wrote: > >> The logic all seems correct here, but I think we can still do better on the >> message. The class and protocol cases seem a little underinformative to me, >> especially since the analyzer doesn't ha

r194086 - Implement AArch64 Neon Crypto instruction classes AES, SHA, and 3 SHA.

Author: jiangning Date: Tue Nov 5 11:42:24 2013 New Revision: 194086 URL: http://llvm.org/viewvc/llvm-project?rev=194086&view=rev Log: Implement AArch64 Neon Crypto instruction classes AES, SHA, and 3 SHA. Added: cfe/trunk/test/CodeGen/aarch64-neon-crypto.c Modified: cfe/trunk/include/cl

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Mon, Nov 4, 2013 at 6:37 PM, Richard Smith wrote: > > On 4 Nov 2013 11:05, "Kaelyn Uhrain" wrote: > > > > Author: rikka > > Date: Mon Nov 4 12:59:34 2013 > > New Revision: 194002 > > > > URL: http://llvm.org/viewvc/llvm-project?rev=194002&view=rev > > Log: > > Try to correct a mistyped "-" o

Re: [PATCH] Implement aarch64 neon instruction class SIMD Bitwise Extract - LLVM

Tim, Sorry, what name do you want to change to? I don't quite understand your comments here. Thanks, -Jiangning > > > Comment at: lib/Target/AArch64/AArch64InstrNEON.td:3953 > @@ +3952,3 @@ > + let ParserMatchClass = uimm3_asmoperand; > + let PrintMethod = "printNeonUImm8Oper

Re: [PATCH] Implement aarch64 neon instruction class SIMD Perm - LLVM

PING... 2013/10/31 Jiangning Liu > Hi t.p.northover, > > http://llvm-reviews.chandlerc.com/D2076 > > Files: > lib/Target/AArch64/AArch64ISelLowering.cpp > lib/Target/AArch64/AArch64InstrFormats.td > lib/Target/AArch64/AArch64InstrNEON.td > test/CodeGen/AArch64/neon-perm.ll > test/MC/A

Re: [PATCH] Fix PR17738 - add support for vtordisp thunks when using -cxx-abi microsoft

Addressed the comments Hi rnk, http://llvm-reviews.chandlerc.com/D2079 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D2079?vs=5293&id=5350#toc Files: include/clang/Basic/ABI.h lib/AST/ItaniumMangle.cpp lib/AST/MicrosoftMangle.cpp lib/AST/VTableBuilder.cpp lib/CodeGen/It

Re: [PATCH] Fix PR17738 - add support for vtordisp thunks when using -cxx-abi microsoft

Please take another look Comment at: include/clang/Basic/ABI.h:111 @@ +110,3 @@ + /// adjustment, if needed. + union VirtualAdjustment { +// Itanium ABI Reid Kleckner wrote: > I'm still not really totally comfortable with the unionification of this > st

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Tue, Nov 5, 2013 at 9:56 AM, Kaelyn Uhrain wrote: > On Mon, Nov 4, 2013 at 6:37 PM, Richard Smith wrote: > >> >> On 4 Nov 2013 11:05, "Kaelyn Uhrain" wrote: >> > >> > Author: rikka >> > Date: Mon Nov 4 12:59:34 2013 >> > New Revision: 194002 >> > >> > URL: http://llvm.org/viewvc/llvm-project

Re: [PATCH] Fix PR17738 - add support for vtordisp thunks when using -cxx-abi microsoft

Fix a test expectation for the Release build Hi rnk, http://llvm-reviews.chandlerc.com/D2079 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D2079?vs=5350&id=5352#toc Files: include/clang/Basic/ABI.h lib/AST/ItaniumMangle.cpp lib/AST/MicrosoftMangle.cpp lib/AST/VTableBuilde

r194088 - Revert "Try to correct a mistyped "-" or ">" to "->" for some C++ cases."

Author: rikka Date: Tue Nov 5 12:28:21 2013 New Revision: 194088 URL: http://llvm.org/viewvc/llvm-project?rev=194088&view=rev Log: Revert "Try to correct a mistyped "-" or ">" to "->" for some C++ cases." Revert this patch until cases of rejected valid code (e.g. identifiers that require ADL to

Re: [PATCH] C++11: Reject string literal to non-const char * conversion

Compared to the previous revision, this revision issues a new ExtWarn instead of the existing Warning. Hi rsmith, http://llvm-reviews.chandlerc.com/D1965 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D1965?vs=5328&id=5351#toc Files: include/clang/Basic/DiagnosticSemaKinds.td

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Tue, Nov 5, 2013 at 10:21 AM, Richard Smith wrote: > On Tue, Nov 5, 2013 at 9:56 AM, Kaelyn Uhrain wrote: > >> On Mon, Nov 4, 2013 at 6:37 PM, Richard Smith wrote: >> >>> >>> On 4 Nov 2013 11:05, "Kaelyn Uhrain" wrote: >>> > >>> > Author: rikka >>> > Date: Mon Nov 4 12:59:34 2013 >>> > New R

Re: r194000 - Use aliases for more constructors and destructors.

The program is rather large, using clang to compile significant C++ code to IR code and the JIT to execute. I do not know enough to make a small testcase. Sorry. Anyhow, the problem started in r194000, in r193999 it's OK. In r194046 (your patch) the problem is indeed solved. With latest ToT r19408

Re: [clang-tools-extra] r194079 - [extra] pp-trace - Fixed problems found while writing docs.

On Tue, Nov 5, 2013 at 10:59 AM, Thompson, John < john_thomp...@playstation.sony.com> wrote: > More tests are coming. I’m doing it incrementally, in groups of related > callbacks. > It's really helpful to commit the tests along with the code change that caused the test to pass - this means that

RE: [clang-tools-extra] r194079 - [extra] pp-trace - Fixed problems found while writing docs.

More tests are coming. I'm doing it incrementally, in groups of related callbacks. From: cfe-commits-boun...@cs.uiuc.edu [mailto:cfe-commits-boun...@cs.uiuc.edu] On Behalf Of David Blaikie Sent: Tuesday, November 05, 2013 8:19 AM To: John Thompson Cc: llvm cfe Subject: Re: [clang-tools-extra] r

Re: [PATCH] Introduce BoundNodes::getMap.

lg http://llvm-reviews.chandlerc.com/D2095 BRANCH clangquery1 ARCANIST PROJECT clang ___ cfe-commits mailing list cfe-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Re: [PATCH] Fix PR17738 - add support for vtordisp thunks when using -cxx-abi microsoft

LGTM Comment at: lib/AST/MicrosoftMangle.cpp:1849 @@ -1848,2 +1848,3 @@ MicrosoftCXXNameMangler &Mangler, raw_ostream &Out) { + if (!Adjustment.Virtual.isEmpty()) { Timur Iskhodzhan

Re: [PATCH] Fix PR17738 - add support for vtordisp thunks when using -cxx-abi microsoft

2013/11/5 Reid Kleckner : > > LGTM > > > > Comment at: lib/AST/MicrosoftMangle.cpp:1849 > @@ -1848,2 +1848,3 @@ >MicrosoftCXXNameMangler &Mangler, >raw_ostream &Out) { > + if (!Adjustment.Virtual.isE

Re: [PATCH] Fix vbtable indices if a class shares the vbptr with a non-virtual base

Comment at: lib/AST/VTableBuilder.cpp:3314-3316 @@ +3313,5 @@ +const ASTRecordLayout &BaseLayout = Context.getASTRecordLayout(CurBase); +if (!BaseLayout.hasVBPtr() || +DerivedVBPtrOffset != BaseOffset + BaseLayout.getVBPtrOffset()) + break; +

r194090 - clang-format: Allow line merging and partial formatting of nested blocks

Author: djasper Date: Tue Nov 5 13:10:03 2013 New Revision: 194090 URL: http://llvm.org/viewvc/llvm-project?rev=194090&view=rev Log: clang-format: Allow line merging and partial formatting of nested blocks Before, clang-format would always format entire nested blocks, which can be unwanted e.g.

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

I'm not sure it works with the CMake build system. When I'm working with TOT (and CMake), I usually symlink the libc++/include dir at the right place in the install dir (lib/c++/v1) so clang found it automatically. That said, produced binaries will always use the system libc++ as by default th

Re: [PATCH] Fix PR17738 - add support for vtordisp thunks when using -cxx-abi microsoft

On Tue, Nov 5, 2013 at 11:10 AM, Timur Iskhodzhanov wrote: > 2013/11/5 Reid Kleckner : > > > > LGTM > > > > > > > > Comment at: lib/AST/MicrosoftMangle.cpp:1849 > > @@ -1848,2 +1848,3 @@ > >MicrosoftCXXNameMangler &Mangler, > >

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Tue, Nov 5, 2013 at 10:40 AM, Kaelyn Uhrain wrote: > On Tue, Nov 5, 2013 at 10:21 AM, Richard Smith wrote: > >> On Tue, Nov 5, 2013 at 9:56 AM, Kaelyn Uhrain wrote: >> >>> On Mon, Nov 4, 2013 at 6:37 PM, Richard Smith wrote: >>> On 4 Nov 2013 11:05, "Kaelyn Uhrain" wrote: > >>

r194092 - Documentation tweaks: Fix a typo in a reference to the C++ standard,

Author: jdennett Date: Tue Nov 5 13:37:38 2013 New Revision: 194092 URL: http://llvm.org/viewvc/llvm-project?rev=194092&view=rev Log: Documentation tweaks: Fix a typo in a reference to the C++ standard, another typo in a comment, and update formatting of some comments to use Doxygen \brief annota

Re: [PATCH] Introduce ClangTool::buildASTs, and buildASTFromCode.

Comment at: unittests/Tooling/ToolingTest.cpp:154 @@ -129,1 +153,3 @@ + clang::tooling::ToolInvocation Invocation(Args, new SyntaxOnlyAction, +&*Files); Invocation.mapVirtualFile("test.cpp", "#include \n"); .getPtr(

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Tue, Nov 5, 2013 at 11:37 AM, Richard Smith wrote: > On Tue, Nov 5, 2013 at 10:40 AM, Kaelyn Uhrain wrote: > >> On Tue, Nov 5, 2013 at 10:21 AM, Richard Smith wrote: >> >>> On Tue, Nov 5, 2013 at 9:56 AM, Kaelyn Uhrain wrote: >>> On Mon, Nov 4, 2013 at 6:37 PM, Richard Smith wrote:

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Tue, Nov 5, 2013 at 11:54 AM, Kaelyn Uhrain wrote: > > > > On Tue, Nov 5, 2013 at 11:37 AM, Richard Smith wrote: > >> On Tue, Nov 5, 2013 at 10:40 AM, Kaelyn Uhrain wrote: >> >>> On Tue, Nov 5, 2013 at 10:21 AM, Richard Smith wrote: >>> On Tue, Nov 5, 2013 at 9:56 AM, Kaelyn Uhrain wrot

Re: r194002 - Try to correct a mistyped "-" or ">" to "->" for some C++ cases.

On Tue, Nov 5, 2013 at 12:14 PM, Richard Smith wrote: > On Tue, Nov 5, 2013 at 11:54 AM, Kaelyn Uhrain wrote: > >> >> >> >> On Tue, Nov 5, 2013 at 11:37 AM, Richard Smith wrote: >> >>> On Tue, Nov 5, 2013 at 10:40 AM, Kaelyn Uhrain wrote: >>> On Tue, Nov 5, 2013 at 10:21 AM, Richard Smith

Re: Produce direct calls instead of alias to linkonce_odr functions

Now with the patch actually attached. On 5 November 2013 00:33, Rafael Espíndola wrote: >>> What happens is that at the end of GetOrCreateLLVMFunction we check if >>> we have a decl defined in class and if so add it to >>> DeferredDeclsToEmit. The net result is that we end up running >>> replace

Re: r194000 - Use aliases for more constructors and destructors.

On 5 November 2013 10:42, Yaron Keren wrote: > The program is rather large, using clang to compile significant C++ code to > IR code and the JIT to execute. I do not know enough to make a small > testcase. Sorry. > > Anyhow, the problem started in r194000, in r193999 it's OK. > In r194046 (your pa

Re: [PATCH] C++11: Reject string literal to non-const char * conversion

Comment at: include/clang/Basic/DiagnosticSemaKinds.td:4488-4489 @@ -4487,2 +4487,4 @@ "conversion from string literal to %0 is deprecated">, InGroup; +def ext_warn_deprecated_string_literal_conversion : ExtWarn< + "conversion from string literal to %0 is ill-formed in C++11

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

The libc++ dylib is shipped with OS X, but the libc++ headers are supposed to go with clang. I don't use CMake but it sounds like someone should fix it to install the headers. On Nov 5, 2013, at 11:23 AM, Jean-Daniel Dupas wrote: > I'm not sure it works with the CMake build system. > > When

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

On Tue, Nov 5, 2013 at 9:44 AM, Bob Wilson wrote: > On OS X the libc++ headers are meant to be installed along with clang. If > you check out the libc++ sources in the LLVM projects directory, they > should be installed in the right place. > Thanks, that makes sense. Should http://clang.llvm.or

Re: [patch] Slightly less hackish implementation of -Wwrite-strings

LGTM On Tue, Oct 29, 2013 at 7:48 AM, Rafael Espíndola < rafael.espind...@gmail.com> wrote: > ping > > On 25 September 2013 17:44, Rafael Espíndola > wrote: > > With this patch -Wwrite-strings is still implemented with the terrible > > hack of passing -fconst-strings to -cc1, but at least the d

Re: Produce direct calls instead of alias to linkonce_odr functions

Looks good, I'm fairly confident this will DTRT. As a followup someone should remove the hack at the top of GetAddrOfCXXDestructor. + typedef llvm::StringMap > ReplacemensType; + ReplacemensType Replacements; typo Do we prefer Type or Ty for typedefs? +void CodeGenModule::applyReplacements()

Re: Produce direct calls instead of alias to linkonce_odr functions

On 5 November 2013 13:23, Reid Kleckner wrote: > Looks good, I'm fairly confident this will DTRT. > > As a followup someone should remove the hack at the top of > GetAddrOfCXXDestructor. > > + typedef llvm::StringMap > > ReplacemensType; > + ReplacemensType Replacements; > > typo > > Do we prefe

r194095 - Produce direct calls instead of alias to linkonce_odr functions.

Author: rafael Date: Tue Nov 5 15:37:29 2013 New Revision: 194095 URL: http://llvm.org/viewvc/llvm-project?rev=194095&view=rev Log: Produce direct calls instead of alias to linkonce_odr functions. This is a small optimization on linux, but should help more on windows where msvc only outputs one

r194097 - With this patch -Wwrite-strings is still implemented with the terrible

Author: rafael Date: Tue Nov 5 15:43:54 2013 New Revision: 194097 URL: http://llvm.org/viewvc/llvm-project?rev=194097&view=rev Log: With this patch -Wwrite-strings is still implemented with the terrible hack of passing -fconst-strings to -cc1, but at least the driver uses the regular warning chec

Re: r193934 - Default to use libc++ on OS X 10.9+ and iOS 7+.

It looks like the CMake build system install the headers in include/c++/v1 while clang expects them in lib/c++/v1 (at least on Darwin). Should it be changed for all targets or should we add a special case for Darwin ? Le 5 nov. 2013 à 22:03, Bob Wilson a écrit : > The libc++ dylib is shipped

r194098 - Refactor constant expression handling and make a couple of tweaks to make it a

Author: rsmith Date: Tue Nov 5 16:18:15 2013 New Revision: 194098 URL: http://llvm.org/viewvc/llvm-project?rev=194098&view=rev Log: Refactor constant expression handling and make a couple of tweaks to make it a bit more robust against future changes. This includes a slight diagnostic improvement:

r194099 - Simplify: we don't care why constant evaluation might have failed when we're

Author: rsmith Date: Tue Nov 5 16:23:30 2013 New Revision: 194099 URL: http://llvm.org/viewvc/llvm-project?rev=194099&view=rev Log: Simplify: we don't care why constant evaluation might have failed when we're checking an expression for constant overflow. Modified: cfe/trunk/include/clang/AST

r194100 - ObjectiveC migrator. Please annotation of properties with

Author: fjahanian Date: Tue Nov 5 16:28:30 2013 New Revision: 194100 URL: http://llvm.org/viewvc/llvm-project?rev=194100&view=rev Log: ObjectiveC migrator. Please annotation of properties with NS_RETURNS_INNER_POINTER under -objcmt-returns-innerpointer-property flag (off by default), as older co

Re: r194090 - clang-format: Allow line merging and partial formatting of nested blocks

On Nov 5, 2013, at 2:24 PM, Daniel Jasper wrote: > Is there more detailed output? Somehow, buildbot went green again. Sorry for the noise. If it happens again, I dig deeper. - Fariborz ___ cfe-commits mailing list cfe-commits@cs.uiuc.edu http://li

[PATCH] Microsoft Bitfields and VBases

Hi majnemer, rsmith, Microsoft adds weird padding before virtual bases if the last field was a bitfield. http://llvm-reviews.chandlerc.com/D2103 Files: lib/AST/RecordLayoutBuilder.cpp test/Layout/ms-x86-bitfields-vbases.cpp Index: lib/AST/RecordLayoutBuilder.cpp ===

Re: [PATCH] Microsoft Bitfields and VBases

This patch fixes the bug: http://llvm.org/bugs/show_bug.cgi?id=17761 http://llvm-reviews.chandlerc.com/D2103 ___ cfe-commits mailing list cfe-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Re: [PATCH] Microsoft Bitfields and VBases

Oh wow. That's ... weird. Patch LGTM. Comment at: test/Layout/ms-x86-bitfields-vbases.cpp:61 @@ +60,3 @@ + +struct D : virtual B0 { char a : 1; char b; }; + What happens for: struct E : virtual B0 { char a, b, c, d : 4, e : 4, : 0; }; ? Do we get extra pad

Re: [PATCH] Microsoft Bitfields and VBases

On Tue, Nov 5, 2013 at 3:13 PM, Richard Smith wrote: > > Oh wow. That's ... weird. Patch LGTM. > > > > Comment at: test/Layout/ms-x86-bitfields-vbases.cpp:61 > @@ +60,3 @@ > + > +struct D : virtual B0 { char a : 1; char b; }; > + > > What happens for: > > str

r194103 - Thread safety analysis: check pt_guarded_by attribute when calling -> and *

Author: delesley Date: Tue Nov 5 17:09:56 2013 New Revision: 194103 URL: http://llvm.org/viewvc/llvm-project?rev=194103&view=rev Log: Thread safety analysis: check pt_guarded_by attribute when calling -> and * on smart pointers. -Wthread-safety-beta only. Modified: cfe/trunk/lib/Analysis/Th

Re: [PATCH] Introduce ClangTool::buildASTs, and buildASTFromCode.

Comment at: include/clang/Tooling/Tooling.h:275 @@ -225,1 +274,3 @@ + /// append them to ASTs. + int buildASTs(std::vector &ASTs); Manuel Klimek wrote: > Peter Collingbourne wrote: > > Manuel Klimek wrote: > > > Hm. ClangTool is made to be extended, for examp

Re: [PATCH] Introduce ClangTool::buildASTs, and buildASTFromCode.

- Use getPtr Hi klimek, http://llvm-reviews.chandlerc.com/D2097 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D2097?vs=5347&id=5359#toc Files: include/clang/Tooling/Tooling.h lib/Tooling/Tooling.cpp unittests/Tooling/ToolingTest.cpp Index: include/clang/Tooling/Tooling.h

Re: [PATCH] Introduce ClangTool::buildASTs, and buildASTFromCode.

- De-virtualize run() Hi klimek, http://llvm-reviews.chandlerc.com/D2097 CHANGE SINCE LAST DIFF http://llvm-reviews.chandlerc.com/D2097?vs=5359&id=5360#toc Files: include/clang/Tooling/Tooling.h lib/Tooling/Tooling.cpp unittests/Tooling/ToolingTest.cpp Index: include/clang/Tooling/T

[PATCH] [-cxx-abi microsoft] Emit thunks for pointers to virtual member functions

Hi rnk, timu, This patch adds support for emitting thunks for pointers to virtual member functions. Unfortunately it doesn't work for variadic functions or functions where the return type or one of the paramters is incomplete, but I figure we should support this broad case first and then f

r194113 - Introduce DynTypedNode::print, dump and getSourceRange.

Author: pcc Date: Tue Nov 5 18:27:12 2013 New Revision: 194113 URL: http://llvm.org/viewvc/llvm-project?rev=194113&view=rev Log: Introduce DynTypedNode::print, dump and getSourceRange. These functions can generally be applied to multiple kinds of AST node, so it makes sense to add them to DynTyp

r194112 - Introduce BoundNodes::getMap.

Author: pcc Date: Tue Nov 5 18:27:07 2013 New Revision: 194112 URL: http://llvm.org/viewvc/llvm-project?rev=194112&view=rev Log: Introduce BoundNodes::getMap. The purpose of this function is to allow clients of the dynamic AST matcher to enumerate each binding. Differential Revision: http://llv

Re: [PATCH] Introduce BoundNodes::getMap.

Closed by commit rL194112 (authored by @pcc). CHANGED PRIOR TO COMMIT http://llvm-reviews.chandlerc.com/D2095?vs=5342&id=5363#toc http://llvm-reviews.chandlerc.com/D2095 COMMIT http://llvm-reviews.chandlerc.com/rL194112 ___ cfe-commits mailing l

Re: [PATCH] Introduce DynTypedNode::print, dump and getSourceRange.

Closed by commit rL194113 (authored by @pcc). CHANGED PRIOR TO COMMIT http://llvm-reviews.chandlerc.com/D2096?vs=5345&id=5364#toc http://llvm-reviews.chandlerc.com/D2096 COMMIT http://llvm-reviews.chandlerc.com/rL194113 ___ cfe-commits mailing l

[PATCH] Microsoft Layout - Empty non-virtual base after base with vbptr

Hi majnemer, rsmith, Microsoft adds an extra byte of padding before laying out zero sized non-virtual bases if the non-virtual base before it contains a vbptr. Addresses: http://llvm.org/bugs/show_bug.cgi?id=17750 http://llvm-reviews.chandlerc.com/D2106 Files: lib/AST/RecordLayoutBuilder.cpp

Re: [PATCH] Implement aarch64 neon instruction class SIMD Bitwise Extract - LLVM

> Sorry, what name do you want to change to? I don't quite understand your > comments here. It's the "printNeonUImm8Operand". It seems to be used for all NEON operands now, not just 8-bit ones. Its distinguishing characteristic appears to be that it prints in hex rather than anything to do with NE

Re: [PATCH] Implement aarch64 neon instructio​n class SIMD Perm - LLVM

Hi Jiangning, This looks reasonable to me. Cheers. Tim. http://llvm-reviews.chandlerc.com/D2076 ___ cfe-commits mailing list cfe-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Re: [PATCH] Microsoft Bitfields and VBases

E does not get padding, F does. even: struct G : virtual B0 { char a, b, c, d : 8; } gets padding. On Tue, Nov 5, 2013 at 3:13 PM, Richard Smith wrote: > On Tue, Nov 5, 2013 at 3:13 PM, Richard Smith wrote: > >> >> Oh wow. That's ... weird. Patch LGTM. >> >> >> >> Comment at

Re: [PATCH] Implement aarch64 neon instructio​n class SIMD Perm - Clang

Hi Jiangning, I've got one question (well, stated as a comment) about this Clang stuff. Cheers. Tim. Comment at: lib/CodeGen/CGBuiltin.cpp:2374-2375 @@ -2373,1 +2373,4 @@ // in aarch64-neon-intrinsics.c so far. + case AArch64::BI__builtin_neon_vuzp_v: +return

PATCH: do ?: lowering to selects less wrong

When lowering "cond ? X : Y" we do some safety checks to see whether we can instead emit both X and Y and use a llvm select instruction to choose between them. This code is insufficiently safe, and introducing loads into a program that didn't load is a bad idea. For example, it could be TLS. It cou

  1   2   >