0183b9f | Timur Pocheptsov | 24 March 2014, 10:16:06 UTC | Make it ... hmm... more C++-like? | 24 March 2014, 10:16:46 UTC |
6e89909 | Fons Rademakers | 24 March 2014, 10:03:40 UTC | Qt >= 4.8 supports C++11, so make that the minimum requirement. | 24 March 2014, 10:03:40 UTC |
05877d4 | Pere Mato | 24 March 2014, 08:52:16 UTC | Enable libcxx also for MacOSX 10.8. | 24 March 2014, 08:52:16 UTC |
945eabc | Danilo Piparo | 24 March 2014, 08:32:30 UTC | Fix SPI-491: Clang33 builds of ROOT | 24 March 2014, 08:32:30 UTC |
746ad28 | Timur Pocheptsov | 24 March 2014, 08:12:19 UTC | More. | 24 March 2014, 08:12:19 UTC |
905abcb | Timur Pocheptsov | 24 March 2014, 08:08:32 UTC | Select the right drawable, not a canvas' background pixmap. Hmm. C++11?? mmkkeyy. | 24 March 2014, 08:08:32 UTC |
7374c9b | Axel Naumann | 23 March 2014, 20:22:09 UTC | Also autoload missing vtable (ROOT-6140). | 23 March 2014, 20:22:09 UTC |
b4f75a2 | Axel Naumann | 23 March 2014, 20:08:48 UTC | Add comment explaining (part of) the bootstrap sequencing. | 23 March 2014, 20:08:48 UTC |
e40e29a | Pere Mato | 23 March 2014, 19:26:55 UTC | Enable C++11 by default (CMake). Fix issue with /etc/cling. | 23 March 2014, 19:26:55 UTC |
856da07 | Timur Pocheptsov | 23 March 2014, 11:21:35 UTC | Not sure though - <algorithm> was not included but in any case - TMath::Min instead of std::min should help. | 23 March 2014, 11:21:35 UTC |
b48495c | Timur Pocheptsov | 23 March 2014, 10:31:51 UTC | Fix some c&p induced problems. | 23 March 2014, 10:31:51 UTC |
cd428ca | Timur Pocheptsov | 23 March 2014, 10:26:39 UTC | New GL tutorials demonstrating a linear gradient fill. | 23 March 2014, 10:26:39 UTC |
71d2859 | Timur Pocheptsov | 23 March 2014, 10:12:14 UTC | Linear gradient for gl-pad. | 23 March 2014, 10:12:14 UTC |
596a8bf | Timur Pocheptsov | 23 March 2014, 09:10:58 UTC | Stencil on win32 and X11. | 23 March 2014, 09:10:58 UTC |
58ba7bd | Timur Pocheptsov | 23 March 2014, 08:59:11 UTC | Request a stencil buffer for gl-pad, we need it now. | 23 March 2014, 08:59:11 UTC |
4a555a0 | Wim Lavrijsen | 23 March 2014, 00:53:08 UTC | improve choice of reported exception type for overloads | 23 March 2014, 00:53:08 UTC |
2e3fdc5 | Axel Naumann | 22 March 2014, 16:27:57 UTC | MASSIVE reduction in linked libs. Ouch. | 22 March 2014, 16:27:57 UTC |
dd8abf8 | CristinaCristescu | 21 March 2014, 12:49:08 UTC | Corrections. (ROOT-5702). | 22 March 2014, 15:41:32 UTC |
1c05384 | Axel Naumann | 22 March 2014, 09:40:36 UTC | Silence Qt warnings. | 22 March 2014, 09:41:07 UTC |
100b2f3 | Axel Naumann | 21 March 2014, 16:31:31 UTC | Expose all files included in PCH/PCM to FileManager. Else, FileManager will fopen(), Preprocessor will notice that there is a defined code guard and not enter the source file, thus leaving the file handle unclosed. Now FileEntry will know that a file comes from a PCh/PCM and not open it when asking for its stat. | 22 March 2014, 09:41:07 UTC |
3242db6 | Bertrand Bellenot | 21 March 2014, 21:10:27 UTC | Remove unreferenced local variable | 21 March 2014, 21:10:27 UTC |
6a0f06e | Bertrand Bellenot | 21 March 2014, 20:47:55 UTC | Implement (finally) correct transparency stipple filling on Windows | 21 March 2014, 20:47:55 UTC |
91eb414 | Danilo Piparo | 21 March 2014, 20:43:28 UTC | Relax a bit the 80 cols limit in TMetaUtils it is indeed not a requirement of ROOT and could have in some cases geopardised the clarity of the code. | 21 March 2014, 20:43:28 UTC |
0c88025 | Danilo Piparo | 21 March 2014, 20:33:39 UTC | Remove thread safety solution for the TNormalizedContext it is of course crucial to have a thread safe ROOT. On the other hand it may be premature to impose such an ad hoc solution for this particular problem. An improved design could maybe avoid the need of a synchronisation mechanism altoghether in this case. | 21 March 2014, 20:33:39 UTC |
a9b1b2e | Danilo Piparo | 21 March 2014, 20:33:12 UTC | Avoid polluting the global namespace with typedefs | 21 March 2014, 20:33:12 UTC |
f9922e5 | Danilo Piparo | 21 March 2014, 14:38:02 UTC | Remove ifdefs depending on the usage of c++11 | 21 March 2014, 20:15:47 UTC |
426cc91 | Danilo Piparo | 21 March 2014, 19:53:45 UTC | Fix stressMathCore (broken because of bogus classnames) Rollback also some functionality for the manipulation of template arguments via dict selection file. Only arguments which are not template specialisations can be treated. | 21 March 2014, 19:53:45 UTC |
ecaf8fc | Olivier Couet | 21 March 2014, 16:16:23 UTC | Implement “Move Opaque”, when “can edit histograms” is on. | 21 March 2014, 16:17:25 UTC |
b8a7542 | Philippe Canal | 21 March 2014, 14:48:31 UTC | Preliminary support for template alias. GetPartiallyDesaguredType now desugar template alias but does not yet properly propagate the opaque typedef and thus: template <typename T> using myvector = std::vector<T>; myvector<Double32_t> vd32d; template<class T> using ptr = T*; ptr<Double32_t> p3; Both currently result in the Double32_t being ignored. | 21 March 2014, 15:50:58 UTC |
9043f8e | Philippe Canal | 20 March 2014, 23:27:54 UTC | No longer hide the compiler detail. The partial desugaring (part of normalization) now strips a typedef even when it original from std and points to compiler details (__gnu_cxx, etc.). The cost of this convenience was the fact the same entity could have different 'name' and thus different class even-though they were supposed to be strictly equivalent. i.e. depending on the way to get to the class it could be name vector<int*>::iterator or __gnu_cxx::iterator<int*> or std::iterator<int*>. This different from the case where we have a strictly bidirectional equivalent between the class and its unique typedef (yes, we are talking about you std::string). This also different from the case where there is a semantic difference between the two possible name. I.e. where an opaque typedef is used. For example, vector<Double32_t> and vector<double> and where the 'way' to get to the type[name] is relevant. | 21 March 2014, 15:50:58 UTC |
2583608 | Olivier Couet | 21 March 2014, 15:39:45 UTC | Move opaque | 21 March 2014, 15:39:45 UTC |
47552b1 | Olivier Couet | 21 March 2014, 15:19:09 UTC | Use transparency | 21 March 2014, 15:22:06 UTC |
19d17bd | Gerardo Ganis | 20 March 2014, 15:33:48 UTC | Keep at most 1 old session logs by default | 21 March 2014, 14:08:23 UTC |
00c8e59 | Gerardo Ganis | 20 March 2014, 13:48:24 UTC | Fix way the path name is passed to cache handling functions | 21 March 2014, 14:08:22 UTC |
30d41f1 | Gerardo Ganis | 21 March 2014, 13:14:25 UTC | Port previous patch on MacOsX | 21 March 2014, 14:08:22 UTC |
2c33255 | Gerardo Ganis | 21 March 2014, 11:48:08 UTC | Add support for secondary groups (ROOT-4126) | 21 March 2014, 14:08:22 UTC |
0f663f5 | Danilo Piparo | 21 March 2014, 14:07:36 UTC | 10.9 fix: copy constructor of std::map is explicit there... | 21 March 2014, 14:07:36 UTC |
ead475b | Danilo Piparo | 12 March 2014, 12:27:05 UTC | Implement the "templ arg swallowing": ROOT-5672 (aka Reflex::Selection) The goal of these changes is to allow entities like: myVector<class T, class alloc = myAllocator<T>> to aquire the normalized name myVector<class T> if the selection file / ROOT::Meta::Selection namespace instruct the system to do so, i.e. to "swallow" a number of template arguments if they have a value identical to the default. In order to achieve this goal, several changes had to be developed in different chapters. 1) Consolidation & Simplification - Special property names, like name, iotypename, nArgsToSkip, do not appear anymore as strings in the code, but their values are centralised in TMetaUtils. - Some constructs were expressed using or counting on it C++11. For example initialiser lists, range based loops, return of big objects by value. - Further factorisation and extraction of common routines in TMetaUtils. 2) Interface Changes & Thread Safety: - TInterpreter::RegisterModule: it now takes a vector of string-int pairs in order to be able to attach to template declarations (in the strings) a number of template arguments to keep. - TNormalizedContext: now contains a table of template declarations pointers and integers. This information is used from within the GetNormalizedName routine (which preserve its interface). The read and write access to this table, implemented as a static map (faster than an unordered_map for small sizes - linear VS const time), is thread safe. This is done in order to support simultaneous dictionary loading and name normalization. This is achieved with an extremely lightweight spinlock based on std::atomic_flag. It is expected that this code path, accessing this map, is not systematically followed (for example for non-template classes). A possible improvement could be represented by the usage of an atomic mechanism to protect the map pointer. Indeed, for every bunch of insertions which potentially happen at library load, one could envisage a reallocation of the map and a replacement of the pointer by this new one (many reads, a few writes - the spinlock approach treats reads and writes symmetrically). 3) Dictionary generation: - The Scanner now fills the template class ptr - int map in the TNormalizedContext for the dictionary generation phase. - rootcling now manages the construction in code of the map of fwd declarations - int. This happens concretely, for historical reasons within the TModuleGenerator. Somehow, rootcling manages the migration of the information in TNormalizedContext at dictionaty creation time to the ROOT runtime (as everything else, by the way :-) ) 4) TMetautils routines: - The code added to TMetaUtils is composed by a main routine, KeepNParams, and othe useful but ancillary functions. - KeepNParams is able, recursively, to build a new QualType with the right number of template arguments. To achieve this goal it follows in parallel the template arguments of the actual template instantiation (which are QualTypes) and the parameters of the template (which are decls). It correctly treats also non-type template arguments (integers). It makes sure that, based on the maximum number of arguments to keep, the new type contains the "right" template arguments, for example skipping the ones for which the argument is equal to the default parameter. 5) New Tools and experimentation: - For the newly developed components and for the most recently added, C++11 was used. - Clang-format was used to comply to the ROOT coding rules (at least the ones that can be respected only lexing). 6) Documentation: - All the algorithms are documented. - The new feature is documented in doxy style in the header DictSelectionReader.h | 21 March 2014, 13:25:54 UTC |
8d3f092 | Timur Pocheptsov | 21 March 2014, 10:19:24 UTC | Protect against recursive definitions (for example, data members of a type we're already priting - either static data member or "cross-references"). | 21 March 2014, 10:19:24 UTC |
15681d9 | Axel Naumann | 21 March 2014, 08:00:42 UTC | Suppress format warnings caused by python headers. | 21 March 2014, 08:00:42 UTC |
0e42a57 | Axel Naumann | 20 March 2014, 17:21:07 UTC | Silence warning "register deprecated". | 20 March 2014, 17:21:07 UTC |
cfab8bc | Lorenzo Moneta | 20 March 2014, 15:25:22 UTC | - Fix the HypoTestInverter::Rebuild to re-set the parameter to the same initial value before generating the toys. Apply also several debug improvements - Apply some fixes for ProofLite - have now a deterministic seed making the result reproducible - allow for setting correctly the number of workers when running in ProofLite Improve StandardHypoTestINvMacro for the rebuild. Add an option to specify the configuration used for rebuilding the limit distribution | 20 March 2014, 15:54:15 UTC |
3c0528e | Axel Naumann | 20 March 2014, 13:03:53 UTC | Silence "declaration shadows a variable in namespace std". | 20 March 2014, 15:18:04 UTC |
6eb4380 | Axel Naumann | 20 March 2014, 13:00:08 UTC | Silence warnings issued by Qt /usr/include/qt4/QtCore/qsharedpointer_impl.h:435:17: error: 'register' storage class specifier is deprecated [-Werror,-Wdeprecated-register] register int tmp = o->strongref; ^~~~~~~~~ /usr/include/qt4/QtCore/qsharedpointer_impl.h:595:43: error: field 'd' is uninitialized when used here [-Werror,-Wuninitialized] inline QWeakPointer(X *ptr) : d(ptr ? d->getAndRef(ptr) : 0), value(ptr) | 20 March 2014, 15:18:03 UTC |
414ec2f | Axel Naumann | 20 March 2014, 12:53:38 UTC | fBuf always evals to true. | 20 March 2014, 15:18:03 UTC |
4ae3532 | CristinaCristescu | 17 March 2014, 15:03:54 UTC | Make TEnum derive from TDictionary and AST check optimazation for IsValid. (ROOT-5702). | 20 March 2014, 15:18:03 UTC |
edb256f | Olivier Couet | 20 March 2014, 14:51:01 UTC | Fix http://root.cern.ch/phpBB3/viewtopic.php?f=3&t=17791 | 20 March 2014, 14:51:01 UTC |
706e899 | Timur Pocheptsov | 20 March 2014, 12:49:05 UTC | No need in Update/Modified calls actually. | 20 March 2014, 12:49:23 UTC |
3d4b6e1 | Bertrand Bellenot | 20 March 2014, 11:07:07 UTC | Fix Jira issue #ROOT-6058 Wrong DeclFileName for timespec Use HeaderSearch on the basename of the header file name, to make sure it properly takes the header file from the include path (e.g. not from /usr/include/bits/) | 20 March 2014, 11:07:07 UTC |
2e50e74 | Bertrand Bellenot | 20 March 2014, 10:34:23 UTC | Use -std=gnu++11 flag for cygwin | 20 March 2014, 10:34:23 UTC |
f6672f2 | Pere Mato | 20 March 2014, 09:53:42 UTC | Added the installation of genreflex and rootcint (copies of roocling) for CMake | 20 March 2014, 09:53:42 UTC |
cdd094f | Timur Pocheptsov | 20 March 2014, 09:03:49 UTC | Fix from Anna - call TCanvas::SupportAlpha and test gPad is not null. Thanks to Bertrand for finding this. | 20 March 2014, 09:03:49 UTC |
aa1ee88 | Timur Pocheptsov | 20 March 2014, 08:47:22 UTC | Remove/add includes. | 20 March 2014, 08:47:22 UTC |
c9e6b9a | Timur Pocheptsov | 20 March 2014, 08:45:59 UTC | Cleanup and comments. | 20 March 2014, 08:45:59 UTC |
10e4549 | Philippe Canal | 19 March 2014, 21:36:47 UTC | Ignore warning in external library (dcache) | 19 March 2014, 21:43:31 UTC |
09c5266 | Philippe Canal | 19 March 2014, 21:42:59 UTC | Do not pass on -ffunction-sections. -ffunction-sections breaks the debugger on some platforms ... and does not help libCling at all. | 19 March 2014, 21:43:27 UTC |
9fa2ca0 | Philippe Canal | 19 March 2014, 21:00:51 UTC | suppress warnings | 19 March 2014, 21:05:53 UTC |
4b71c47 | Philippe Canal | 19 March 2014, 20:45:01 UTC | suppress warnings | 19 March 2014, 21:05:38 UTC |
521f394 | Philippe Canal | 19 March 2014, 20:41:01 UTC | fix warnings Conflicts: cint/cint/inc/G__ci.h cint/cint/inc/G__ci_fproto.h | 19 March 2014, 21:04:23 UTC |
9d192c5 | Timur Pocheptsov | 19 March 2014, 20:04:34 UTC | A new demo to ... demonstrate: 1. Linear and radial gradient fills. 2. Shadows. 3. Transparency. + TPie/TText and transparent TPad. | 19 March 2014, 20:06:05 UTC |
87a6392 | Bertrand Bellenot | 19 March 2014, 19:50:04 UTC | Fix compilation error on cygwin/gcc4.8.2 (error: 'to_string' is not a member of 'std') | 19 March 2014, 19:50:04 UTC |
0ec6956 | Bertrand Bellenot | 19 March 2014, 19:38:59 UTC | Add c++11 compilation flags for cygwin/gcc | 19 March 2014, 19:38:59 UTC |
f0df27c | Timur Pocheptsov | 19 March 2014, 18:55:36 UTC | radii | 19 March 2014, 18:55:48 UTC |
34e47d1 | Philippe Canal | 19 March 2014, 18:09:20 UTC | Making warning an error for cling too | 19 March 2014, 18:09:20 UTC |
c646239 | Philippe Canal | 19 March 2014, 18:08:58 UTC | Remove inadervently added -g | 19 March 2014, 18:08:58 UTC |
9530dda | Philippe Canal | 19 March 2014, 17:41:27 UTC | Use enum rather than casting to int | 19 March 2014, 17:49:51 UTC |
7c9b0e9 | Philippe Canal | 19 March 2014, 17:27:25 UTC | Comment out unused typedef | 19 March 2014, 17:49:51 UTC |
75a3238 | Philippe Canal | 19 March 2014, 17:24:37 UTC | Use -Werror but do not propagate it to ACLiC | 19 March 2014, 17:49:50 UTC |
cd02bf8 | Timur Pocheptsov | 19 March 2014, 17:44:31 UTC | One of radii (but not both) can be 0. actually. | 19 March 2014, 17:44:31 UTC |
404cab6 | Timur Pocheptsov | 19 March 2014, 17:16:40 UTC | Fix for a stressSpectrum. | 19 March 2014, 17:17:03 UTC |
b5dfb1d | Pere Mato | 19 March 2014, 16:32:49 UTC | Fix for ROOT-6161. Use macro ROOTICONPATH instead of assuming $ROOTSYS/icons. | 19 March 2014, 16:32:49 UTC |
e657d01 | Fons Rademakers | 19 March 2014, 16:09:07 UTC | enable cocoa back-end by default for OSX >= 10.7. | 19 March 2014, 16:09:07 UTC |
afee5b3 | Olivier Couet | 19 March 2014, 15:40:55 UTC | Ignore the dots spacing if the transparency is available. | 19 March 2014, 15:40:55 UTC |
af67d19 | Timur Pocheptsov | 19 March 2014, 15:02:48 UTC | Add a radial gradient. | 19 March 2014, 15:03:09 UTC |
8a16576 | Bertrand Bellenot | 19 March 2014, 14:47:55 UTC | Add required gcc version | 19 March 2014, 14:47:55 UTC |
73408e5 | Bertrand Bellenot | 19 March 2014, 14:44:06 UTC | re-introduce win32gcc and win64gcc Re-introduce win32gcc and win64gcc and use -std=gnu++11 instead of -std=c++11. This solve undefined snprintf and strdup (at least) on cygwin/gcc4.8-2 | 19 March 2014, 14:44:06 UTC |
f86fea6 | Olivier Couet | 19 March 2014, 14:20:27 UTC | Marker size was wrong in divided canvases. | 19 March 2014, 14:21:13 UTC |
c7254a1 | Timur Pocheptsov | 19 March 2014, 12:15:02 UTC | No need in a specia function "and copy and paste". | 19 March 2014, 12:46:37 UTC |
6dbe999 | Axel Naumann | 19 March 2014, 12:31:20 UTC | Add doc. | 19 March 2014, 12:31:33 UTC |
7cc2010 | Axel Naumann | 19 March 2014, 12:31:09 UTC | Comments help. | 19 March 2014, 12:31:33 UTC |
e3dec6e | Pere Mato | 19 March 2014, 11:40:49 UTC | Fix for ROOT-6158. It adds -iROOTINCDIR in argument before invoking cling interpreter | 19 March 2014, 11:40:49 UTC |
a48697c | Axel Naumann | 19 March 2014, 10:53:54 UTC | Rename for symmetry with hsimpleProxy.C | 19 March 2014, 10:53:57 UTC |
c5017c8 | Olivier Couet | 19 March 2014, 10:25:53 UTC | Show either the Dot Spacing Slider or the opacity one depending if the Canvas support Opacity or not. (From Anna). | 19 March 2014, 10:26:54 UTC |
e4efa2b | Pere Mato | 19 March 2014, 09:43:22 UTC | Introduce a fix that was in the patches branch | 19 March 2014, 09:50:56 UTC |
31d95d4 | Fons Rademakers | 19 March 2014, 09:46:53 UTC | correctly handle deprecated options, like --disable-cxx11. | 19 March 2014, 09:47:36 UTC |
6f651b3 | Timur Pocheptsov | 19 March 2014, 09:12:36 UTC | Make API/interfaces more reasonable and simple. | 19 March 2014, 09:12:36 UTC |
16edbc9 | Fons Rademakers | 18 March 2014, 23:03:19 UTC | correctly place -std=c++11 option. | 18 March 2014, 23:03:19 UTC |
f658148 | Fons Rademakers | 18 March 2014, 22:52:35 UTC | always use flag -std=c++11. | 18 March 2014, 22:52:35 UTC |
9c49bcb | Fons Rademakers | 18 March 2014, 22:31:09 UTC | updated set of supported ROOT 6 architectures. | 18 March 2014, 22:31:39 UTC |
f194a39 | Fons Rademakers | 18 March 2014, 22:07:51 UTC | CINT has been removed. | 18 March 2014, 22:31:39 UTC |
56bb625 | Pere Mato | 18 March 2014, 22:03:10 UTC | Silence some more warnings | 18 March 2014, 22:03:10 UTC |
3061d5c | Philippe Canal | 18 March 2014, 20:30:14 UTC | Avoid order of initialization problem in CINT/Core/Meta. Avoid to forward declare TTree in an header file in core/base as this tickles (on Ubuntu) an order of initialization problems [no point in solving the rarely appearing underlying issue in v5] Import in v6 to keep code in sync. (cherry picked from commit acd8ecba7025d576050186035ffdad82a034aa7c) | 18 March 2014, 21:05:28 UTC |
9345ba9 | Axel Naumann | 18 March 2014, 20:34:59 UTC | Work around ROOT-6016 by skipping timespec. | 18 March 2014, 20:35:13 UTC |
4b77a25 | Timur Pocheptsov | 18 March 2014, 20:06:01 UTC | includes. | 18 March 2014, 20:06:01 UTC |
e5ecef2 | Timur Pocheptsov | 18 March 2014, 19:58:29 UTC | Now Cocoa + Quartz == awesomness can be saved into png/jpg/gif via TASImage as it it. Damn, I'm good. | 18 March 2014, 19:58:29 UTC |
95d40d8 | Timur Pocheptsov | 18 March 2014, 19:03:19 UTC | Fix (again) a start/end point calculation: must work with Cocoa coords, not ROOT's coordinates. | 18 March 2014, 19:03:19 UTC |
d0d9d06 | Timur Pocheptsov | 18 March 2014, 18:17:39 UTC | Fix a bug in start/end points calculation for object bounding mode. | 18 March 2014, 18:17:39 UTC |
8fd209b | Timur Pocheptsov | 18 March 2014, 18:11:06 UTC | Update grad.C tutorial to work with TLinearGradient. | 18 March 2014, 18:11:06 UTC |
4ac4abb | Timur Pocheptsov | 18 March 2014, 18:04:07 UTC | Update a grad2.C tutorial to use TLinearGradient. | 18 March 2014, 18:04:07 UTC |
cf3791a | Timur Pocheptsov | 18 March 2014, 17:48:48 UTC | Make gradient fill more interesting and more real: TRadialGradient and TLinearGradient are two new gradient fill options. | 18 March 2014, 17:49:55 UTC |