93c381c | Axel Naumann | 30 September 2014, 11:49:03 UTC | Set release date. | 30 September 2014, 11:49:03 UTC |
3eea9f7 | Axel Naumann | 30 September 2014, 11:46:44 UTC | Update ROOT version files to v6.02/00. | 30 September 2014, 11:46:44 UTC |
f1f1de3 | Vassil Vassilev | 29 September 2014, 13:56:15 UTC | Add the ASTContext which will provide clang::Attr's operator new. (cherry picked from commit 44498c6c959764b116f3c32cb225b6dc0894ffbe) | 30 September 2014, 05:35:12 UTC |
7111a94 | Vassil Vassilev | 29 September 2014, 13:45:07 UTC | Do not deserialize decls, while iterating to unload. (cherry picked from commit 851be3dca8511ac1117796313ebfe339e1f610aa) | 30 September 2014, 05:35:06 UTC |
ed6bb8a | Vassil Vassilev | 29 September 2014, 13:32:10 UTC | Mark implicit members coming from a PCH as used. Implicitly generated members from a CXXRecordDecls coming from a PCH should not be unloaded. CodeGen 'records' that it once emitted the implicit members and if they get unloaded it would be out of sync. The key difference is the information about all the implicitly generated members (trivial ctors/dtors, etc) comes from an ASTMutationListener and they are not part of the transaction. We still do not record it as a part of the transaction, we just mark them as used so that they don't get unloaded from the llvm::Module. Once the list of llvm.used decls is emitted we need to clear it, otherwise we end up emitting one and the same thing over and over. Fixes ROOT-6722 Tests should be coming shortly. (cherry picked from commit 2abc4e82da4b52c3b9b703774c8f4d03d8647649) | 30 September 2014, 05:34:59 UTC |
2755fee | Danilo Piparo | 29 September 2014, 20:02:30 UTC | Add memory stats before and after autoparsing w/ debug > 0 | 29 September 2014, 20:03:11 UTC |
2370baa | Axel Naumann | 29 September 2014, 15:57:52 UTC | From Danilo and me: add missing member init (fixes CMS). | 29 September 2014, 15:57:52 UTC |
6ff8749 | Axel Naumann | 29 September 2014, 15:39:32 UTC | From Danilo and me: normalize again after resolving a typedef (fixes CMS). | 29 September 2014, 15:39:32 UTC |
8c1ae13 | Lorenzo Moneta | 29 September 2014, 15:19:46 UTC | Set the removal version for the obsolete methods in TH1 to be 6.04 | 29 September 2014, 15:28:17 UTC |
3022046 | Lorenzo Moneta | 25 September 2014, 14:53:39 UTC | Fix a -Wshadow warning | 28 September 2014, 10:24:40 UTC |
4675eae | Lorenzo Moneta | 25 September 2014, 12:49:05 UTC | Add Anderson-Darling test for histograms. See ROOT-6069 | 28 September 2014, 10:24:40 UTC |
c2af1de | Lorenzo Moneta | 25 September 2014, 12:46:15 UTC | Try to fix test for ubuntu. Add drawing of graph as an option | 28 September 2014, 10:24:40 UTC |
707318a | Lorenzo Moneta | 25 September 2014, 10:12:56 UTC | Initialize some variables | 28 September 2014, 10:24:40 UTC |
10d5454 | Lorenzo Moneta | 25 September 2014, 14:17:51 UTC | Disable PROOF for these tutorials | 28 September 2014, 10:24:39 UTC |
85b7cef | Lorenzo Moneta | 26 September 2014, 07:53:43 UTC | Speed up tutorial by plotting the posterior in less points (20 instead of 50) | 28 September 2014, 10:24:39 UTC |
2f2d73d | Lorenzo Moneta | 27 September 2014, 11:45:35 UTC | initialise data member in default constructor | 28 September 2014, 10:24:39 UTC |
f94d287 | Lorenzo Moneta | 26 September 2014, 13:49:37 UTC | From Wouter: new version of RooChebyshev which support the possibility to define a reference range for the evaluation and normalisation of the polynomial. This fixes the problem observed fitting in 2 distinct ranges, ROOT-6664 | 28 September 2014, 10:24:39 UTC |
b682952 | Gerardo Ganis | 27 September 2014, 08:06:03 UTC | Fix for ROOT-6653 | 27 September 2014, 08:45:48 UTC |
91b8527 | Danilo Piparo | 26 September 2014, 16:31:59 UTC | Fix ROOT-6747: enums treatment in TCling::CheckClassInfo an issue was present with the TCling::CheckClassInfo method. It returned true, i.e. name known to the interpreter, even if the name was an enum. This behaviour can now be steered with a boolean that restricts the search to classes/structs and namespaces only. The problem that allowed to identify this unintended behaviour was the following: a TClass instance was deleting itself during its own destruction, leading to a double delete. This inconsistent state was due to the fact that a last-resort operation is performed in TClass::GetClass, a lookup into the interpreter database. The check was the following: //last attempt. Look in CINT list of all (compiled+interpreted) classes if (gInterpreter->CheckClassInfo(normalizedName.c_str()) { // Get the normalized name based on the decl (currently the only way // to get the part to add or drop the default arguments as requested by the user) std::string alternative; gInterpreter->GetInterpreterTypeName(normalizedName.c_str(),alternative,kTRUE); const char *altname = alternative.c_str(); if ( strncmp(altname,"std::",5)==0 ) { // For namespace (for example std::__1), GetInterpreterTypeName does // not strip std::, so we must do it explicitly here. altname += 5; } if (altname != normalizedName && strcmp(altname,name) != 0) { // altname now contains the full name of the class including a possible // namespace if there has been a using namespace statement. // At least in the case C<string [2]> (normalized) vs C<string[2]> (altname) // the TClassEdit normalization and the TMetaUtils normalization leads to // two different space layout. To avoid an infinite recursion, we also // add the test on (altname != name) return GetClass(altname,load); } TClass *ncl = gInterpreter->GenerateTClass(normalizedName.c_str(), /* emulation = */ kFALSE, silent); if (!ncl->IsZombie()) { return ncl; } delete ncl; } The if clause, if normalizedName refers to an enum should not pass in order not to bootstrap a TClass from an enum name. There is of course a test in TClass::GetClass which checks if the name corresponds to a selected enum: // Try to see if this is an enumerator if(TEnum::GetEnum(name,load ? TEnum::kAutoload : TEnum::kNone)) return nullptr; If unfortunately the user did not select the enum and no parsing happened, the name is unknowm to the system and the static method returns nullptr. | 26 September 2014, 17:05:44 UTC |
e792c42 | Axel Naumann | 26 September 2014, 14:02:33 UTC | Handle member fun ptr (ROOT-5467). (cherry picked from commit 20b95e364e969bdbd64a62dc7ae24a170570732e) | 26 September 2014, 14:03:14 UTC |
f3f5527 | Olivier Couet | 26 September 2014, 13:54:00 UTC | Add a protection to avoid the crash mentionned here: https://sft.its.cern.ch/jira/browse/ROOT-6675 | 26 September 2014, 14:01:16 UTC |
ef1b502 | Olivier Couet | 26 September 2014, 14:00:27 UTC | Fix in case for zoom box in log scale. | 26 September 2014, 14:00:27 UTC |
78ad358 | David Smith | 26 September 2014, 13:32:14 UTC | Do not auto replace explicitly set, default sized TTreeCache (ROOT-6698). | 26 September 2014, 13:32:14 UTC |
6801417 | Axel Naumann | 26 September 2014, 09:57:59 UTC | Use the fully qualified name. (cherry picked from commit accc23e668151c53b8095438efc252b4d429e50e) | 26 September 2014, 10:41:48 UTC |
77eda84 | Axel Naumann | 25 September 2014, 10:31:50 UTC | Update ROOT version files to v6.02/00-rc1. | 25 September 2014, 10:31:50 UTC |
2dc815d | Vassil Vassilev | 25 September 2014, 08:20:49 UTC | Me interacting with MD is always fun. Fix layout and inclusion header. | 25 September 2014, 08:20:49 UTC |
75bed0f | Vassil Vassilev | 25 September 2014, 08:12:33 UTC | Don't forget the new line. | 25 September 2014, 08:12:33 UTC |
89655ab | Vassil Vassilev | 25 September 2014, 07:53:59 UTC | Document that cling driver can be given statements to execute. | 25 September 2014, 07:53:59 UTC |
6fdff10 | Olivier Couet | 25 September 2014, 07:39:11 UTC | More fixes because of new minus | 25 September 2014, 07:39:11 UTC |
3ca2e3a | Wim Lavrijsen | 24 September 2014, 18:54:34 UTC | for Thomas Ruf: handle return of TClonesArray in TTreeGetAttr | 24 September 2014, 18:54:34 UTC |
75d6e04 | Bertrand Bellenot | 24 September 2014, 18:34:04 UTC | Fix Jira issue #ROOT-6680 Display issues with TGTextEntry Fix also an issue with partially highlighted text not moving when resizing the text entry | 24 September 2014, 18:34:04 UTC |
7c8cba8 | Olivier Couet | 24 September 2014, 16:44:04 UTC | The minus should not be changes in time axis labels. | 24 September 2014, 16:44:04 UTC |
aac2d6e | Olivier Couet | 24 September 2014, 15:29:25 UTC | Mods needed because of the new minus sign on axis. | 24 September 2014, 15:29:25 UTC |
f16ed85 | Olivier Couet | 24 September 2014, 13:45:01 UTC | New drawing option "violin" for 2D histograms. See https://sft.its.cern.ch/jira/browse/ROOT-6681 | 24 September 2014, 13:45:01 UTC |
5ceb599 | Olivier Couet | 24 September 2014, 13:08:01 UTC | Fix a typo "to to" should be "to" | 24 September 2014, 13:08:01 UTC |
89bd5e8 | Danilo Piparo | 24 September 2014, 12:52:26 UTC | Turn off association of headers to pcms: not needed by ROOT pcms and problematic for CMS and Atlas: ROOT-6739 | 24 September 2014, 12:53:29 UTC |
0c56caf | Pere Mato | 23 September 2014, 14:21:32 UTC | Added support for COMPONENTS in FindROOT.cmake (ROOT-6717) | 24 September 2014, 12:26:22 UTC |
28bfa81 | Axel Naumann | 24 September 2014, 09:23:19 UTC | Find the RedeclTmplt with the default args. | 24 September 2014, 09:23:19 UTC |
b75acec | Axel Naumann | 24 September 2014, 09:22:45 UTC | Use the canonical decl as keep-n-args key; we might have had a previous decl. | 24 September 2014, 09:22:45 UTC |
e758037 | Axel Naumann | 24 September 2014, 08:36:21 UTC | Remove (now) unused code. | 24 September 2014, 08:36:21 UTC |
8847cb3 | Axel Naumann | 24 September 2014, 08:35:58 UTC | For template instances, forward declare its arguments. Still missing a fwd decl of explicit instantiations! | 24 September 2014, 08:35:58 UTC |
40dbc7c | Axel Naumann | 24 September 2014, 08:35:15 UTC | Use cling fwd decl for typedefs for consistency (ROOT-6595). | 24 September 2014, 08:35:15 UTC |
c837d26 | Axel Naumann | 23 September 2014, 18:48:42 UTC | Declare outermost namespace first. | 23 September 2014, 18:49:12 UTC |
e015874 | Axel Naumann | 23 September 2014, 14:42:03 UTC | Decls as template params need to be buffered to their stream. | 23 September 2014, 18:49:12 UTC |
a13d5cb | Axel Naumann | 23 September 2014, 14:33:01 UTC | Write declcontext for each visit; delay template <...>. Writing the template parameters might cause visitation of other decls which need to be in their respective namespaces. Thus we need to buffer / delay the whole template decl, including its enclosing namespaces. As we forward e.g. from ClassTemplateInstanceDecl() to RedeclarableTemplateDecl() we need to selectively write the enclosing namespaces, no just for any Visit(Decl*). Make sure that a template pattern is not again writing its enclosing scopes - it is just called from its VisitRedeclarableTemplateDecl(). | 23 September 2014, 18:49:12 UTC |
6981e12 | Axel Naumann | 23 September 2014, 13:17:04 UTC | Remove skip counting; centralize skip logging. | 23 September 2014, 18:49:12 UTC |
a7e78c4 | Axel Naumann | 23 September 2014, 13:14:00 UTC | Write to Out() not Log()! | 23 September 2014, 18:49:12 UTC |
657ad07 | Axel Naumann | 23 September 2014, 12:47:18 UTC | Actually print the enclosing DeclContexts. | 23 September 2014, 18:49:12 UTC |
8e983a8 | Axel Naumann | 23 September 2014, 12:37:05 UTC | extern "C" is printed by the decl itself. | 23 September 2014, 18:49:12 UTC |
ec92658 | Axel Naumann | 23 September 2014, 12:35:33 UTC | Let decls print their decl context. | 23 September 2014, 18:49:12 UTC |
cab3e31 | Axel Naumann | 23 September 2014, 12:34:27 UTC | Delay template parameters until its dependencies have been visited. | 23 September 2014, 18:49:12 UTC |
b3d1e5f | Axel Naumann | 23 September 2014, 12:32:53 UTC | No need for a sub-stream; we reset m_SkipFlag. | 23 September 2014, 18:49:11 UTC |
50bcf06 | Axel Naumann | 23 September 2014, 12:31:25 UTC | Automatically stream to parent if not skipped. | 23 September 2014, 18:49:11 UTC |
824bd5a | Axel Naumann | 23 September 2014, 12:29:25 UTC | Fix fatal spelling error. | 23 September 2014, 18:49:11 UTC |
5f0b6df | Axel Naumann | 23 September 2014, 10:05:20 UTC | Even fwd decl outside TU context. Visit tmplt arg types. | 23 September 2014, 18:49:11 UTC |
0a076c5 | Axel Naumann | 23 September 2014, 10:04:36 UTC | Disable fwd decl dependency analysis from TMetaUtil; pulls in the world. | 23 September 2014, 18:49:11 UTC |
45b3b71 | Olivier Couet | 23 September 2014, 13:56:23 UTC | Implement what is suggested here: https://sft.its.cern.ch/jira/browse/ROOT-6711 | 23 September 2014, 13:56:23 UTC |
34e10f6 | Bertrand Bellenot | 23 September 2014, 10:24:58 UTC | From Oliver Freyermuth: Fix a small issue in TSQLiteStatement Fix Jira issue #ROOT-6511 TSQLiteStatement: Missing / wrong check for column index, failing IsNull(), as reported by Oliver Freyermuth: Finally, my colleagues managed to uncover a small issue in my TSQLiteStatement implementation: "IsNull()" checked the column-count in the wrong way (it used the number of bound columns which is only valid in prepared inserts). I have done the following changes to TSQLiteStatement in the attached patch: 1. In fNumPars, I now store either the "bind" columns for a prepared statement with bound parameters, or the number of returned columns for a simple select without binding. This also simplifies error handling and TSQLiteStatement::GetNumFields() now just returns the variable which is correct in both working modes. 2. All Get'ters now check the column-number consistently, as the SQLite-documentation states that "get"ting an invalid column number returns undefined results. Note that for bound parameters, SQLite behaves differently and does the error handling itself (which I already check in TSQLiteStatement). | 23 September 2014, 10:24:58 UTC |
0e30aec | Axel Naumann | 22 September 2014, 19:23:58 UTC | Degenerate floats cast to int can do amazing things. | 22 September 2014, 19:23:58 UTC |
4402826 | Axel Naumann | 22 September 2014, 13:52:38 UTC | Do not use (non-0-term) data(); factor out the tag. | 22 September 2014, 14:30:15 UTC |
27df8a6 | Axel Naumann | 22 September 2014, 13:51:59 UTC | Introduce ASTContext member. | 22 September 2014, 14:30:15 UTC |
4d2b3f7 | Axel Naumann | 22 September 2014, 13:04:01 UTC | Desugar to simplify the type. | 22 September 2014, 14:30:15 UTC |
e45ab6d | Axel Naumann | 19 September 2014, 14:54:23 UTC | Skip if needed when visiting template args. | 22 September 2014, 14:30:15 UTC |
72c1d04 | Axel Naumann | 19 September 2014, 14:54:04 UTC | Do not stream const vars; for static ones, do not print "extern". | 22 September 2014, 14:30:15 UTC |
3d7af7e | Axel Naumann | 19 September 2014, 14:53:22 UTC | Handle DeclRefExp template args. | 22 September 2014, 14:30:15 UTC |
2ae0fc6 | Axel Naumann | 19 September 2014, 14:51:17 UTC | Make sure we have a cling autoload annotation. | 22 September 2014, 14:30:15 UTC |
7b16d00 | Danilo Piparo | 22 September 2014, 13:52:09 UTC | Coding conventions | 22 September 2014, 13:52:09 UTC |
d11d5a0 | Danilo Piparo | 22 September 2014, 13:45:25 UTC | We do not need headers of Xrd in the interface of TNetXNGFile | 22 September 2014, 13:45:25 UTC |
08b0841 | Pere Mato | 21 September 2014, 17:27:30 UTC | Fix for ROOT-6728 - Compilation error in TTimeStamp | 21 September 2014, 17:27:43 UTC |
4622d42 | Vassil Vassilev | 20 September 2014, 19:50:28 UTC | Spelling. | 20 September 2014, 19:50:28 UTC |
9a97d32 | Vassil Vassilev | 20 September 2014, 19:32:39 UTC | Do not skip comments in strings. Fixes ROOT-6692. | 20 September 2014, 19:32:39 UTC |
9e5368a | Pere Mato | 20 September 2014, 14:19:05 UTC | Finish - Do not display suggestions only for ROOT. | 20 September 2014, 14:19:05 UTC |
468652e | Pere Mato | 19 September 2014, 21:14:58 UTC | Ensure that the order of output is respected in tests. | 19 September 2014, 21:14:58 UTC |
7dc9a77 | Danilo Piparo | 19 September 2014, 15:41:24 UTC | Coding conventions. | 19 September 2014, 15:41:43 UTC |
37eefe6 | Danilo Piparo | 19 September 2014, 15:40:39 UTC | Fix ROOT-6724: remove header inclusions in favour of fwd decls. | 19 September 2014, 15:41:43 UTC |
2caa9f2 | Olivier Couet | 19 September 2014, 13:13:01 UTC | Fix: https://sft.its.cern.ch/jira/browse/ROOT-6658 Thanks Bertrand. | 19 September 2014, 13:13:01 UTC |
e44b877 | Vassil Vassilev | 19 September 2014, 11:59:31 UTC | Do not display suggestions only for ROOT. | 19 September 2014, 11:59:35 UTC |
bb66b9a | Philippe Canal | 19 September 2014, 10:38:25 UTC | Don't const_cast to ... const | 19 September 2014, 10:38:25 UTC |
3a1c58d | Axel Naumann | 19 September 2014, 09:11:06 UTC | Write function fwd decls! Follow DeclRefExpr template args. | 19 September 2014, 09:11:14 UTC |
3cb74d6 | Axel Naumann | 18 September 2014, 22:12:17 UTC | Silence warning. | 18 September 2014, 22:12:17 UTC |
d951685 | Axel Naumann | 18 September 2014, 21:43:48 UTC | Fix position of "inline" in "inline namespace" (MacOS G__MathCore failure). | 18 September 2014, 21:43:48 UTC |
e5f0998 | Axel Naumann | 18 September 2014, 21:40:14 UTC | Silence enum typed/untyped redecl until fwd decl can handle it. | 18 September 2014, 21:40:14 UTC |
a67ad14 | Axel Naumann | 18 September 2014, 21:22:02 UTC | Fix mem corruption: iterator might be invalidated by insertion. | 18 September 2014, 21:22:02 UTC |
b0df2b8 | Axel Naumann | 18 September 2014, 20:59:22 UTC | Analyze and fwd declare all dependencies (ROOT-6705 ROOT-6719 ROOT-6712 ROOT-6695). | 18 September 2014, 21:00:28 UTC |
00697c8 | Axel Naumann | 16 September 2014, 11:45:30 UTC | Also fwd declare types nested in namespaces. | 18 September 2014, 21:00:28 UTC |
37f8ee9 | Philippe Canal | 18 September 2014, 18:15:43 UTC | Add note about race condition | 18 September 2014, 20:25:07 UTC |
21181c1 | Philippe Canal | 18 September 2014, 15:37:21 UTC | In CheckSum properly handle the 'signed char' in the Reflex case. gccxml was illegally replacing signed char with char ... so for the checksum equivalency, we need to do the same (only in the emulate-reflex modes) | 18 September 2014, 15:38:21 UTC |
1d63f0f | Bertrand Bellenot | 18 September 2014, 14:50:49 UTC | Make the THttpServer::ProcessRequest() method virtual | 18 September 2014, 14:50:49 UTC |
975324b | Danilo Piparo | 18 September 2014, 14:33:24 UTC | Check for the proxy w/o relying on name which makes the check even more powerful: we now do not check for mother classes not only for stl collections but also for all proxied collections. This is perfectly consistent with the concept of collection proxy, which is there exactly to abstract from the details of the container and focus on the contained entities. | 18 September 2014, 14:35:18 UTC |
2de8f37 | Lorenzo Moneta | 18 September 2014, 13:41:38 UTC | Remove assert. Fix for ROOT-6664 | 18 September 2014, 13:41:38 UTC |
7a9149d | Lorenzo Moneta | 18 September 2014, 09:54:28 UTC | Add a const version of TAxis::FindBin for backward compatibility after having added a const and non-const version of TH1::GetAxis() | 18 September 2014, 09:54:28 UTC |
b6705d8 | Lorenzo Moneta | 18 September 2014, 09:42:27 UTC | Copy from 5.34 fixes for a warning | 18 September 2014, 09:42:43 UTC |
def2a9d | Olivier Couet | 18 September 2014, 09:25:27 UTC | Fix https://sft.its.cern.ch/jira/browse/ROOT-6606 A call to InitializeGraphics is needed in TCanvas:Draw() in case the canvas comes from a file. The previous attempt to fix it in TSystem::Load was wrong. | 18 September 2014, 09:25:27 UTC |
7f3348c | Philippe Canal | 17 September 2014, 21:18:57 UTC | Properly handle the template template args in global scope. Also properly handle the case where the template template argument as a partially qualified scope | 17 September 2014, 23:53:15 UTC |
9c46bfb | Philippe Canal | 17 September 2014, 18:37:11 UTC | Add support for fully qualifying a template argument that is itself a template. This is the case for example with gcc's __gnu_cxx::__common_pool_policy<__gnu_cxx::__pool,true> where __pool is actually a class template. Also reduce (a little bit) code duplication. | 17 September 2014, 23:53:15 UTC |
40df3ca | Philippe Canal | 17 September 2014, 21:05:04 UTC | Better error message | 17 September 2014, 21:05:36 UTC |
ccdf369 | Philippe Canal | 17 September 2014, 21:04:30 UTC | Proper type number for kDataTypeAliasSignedChar_t | 17 September 2014, 21:05:36 UTC |
a71c2a2 | Philippe Canal | 17 September 2014, 21:00:00 UTC | Revert "Add support for fully qualifying a template argument that is itself a template." until we understand the negative side-effects. This reverts commit 19f18e848d50bd44a011e8cf192f56d73e28bf65. | 17 September 2014, 21:00:50 UTC |
74fde3a | Pere Mato | 17 September 2014, 20:58:36 UTC | Align recent changes in patches to master. | 17 September 2014, 20:58:36 UTC |
19f18e8 | Philippe Canal | 17 September 2014, 18:37:11 UTC | Add support for fully qualifying a template argument that is itself a template. This is the case for example with gcc's __gnu_cxx::__common_pool_policy<__gnu_cxx::__pool,true> where __pool is actually a class template. Also reduce (a little bit) code duplication. | 17 September 2014, 18:50:18 UTC |