| 4ecf940 | Lorenzo Moneta | 11 February 2019, 14:26:09 UTC | Add another fix in computation of the shortest interval in the BayesianCalculator. - The computed interval was shifted by one bin. - The reference file did not need to be updates since the reference results are hard coded and taken from a paper. - Add now a new method returning the histogram approximating the posterior. - Increase the number of bins when computing the posterior histogram in stressRooStats | 11 February 2019, 14:26:09 UTC |
| b66d552 | Lorenzo Moneta | 08 February 2019, 15:11:04 UTC | Revert "update refeerence file for changes in Bayesian shortest interval after bug fix" This reverts commit d51c85a5ac1a32161d6e66d10b9426d242591d0c. | 11 February 2019, 14:12:44 UTC |
| d51c85a | Lorenzo Moneta | 08 February 2019, 15:11:04 UTC | update refeerence file for changes in Bayesian shortest interval after bug fix | 08 February 2019, 15:11:04 UTC |
| 059f4d2 | Lorenzo Moneta | 08 February 2019, 14:54:52 UTC | fix a bug in excluding underflow bin when computing shortest interval in BayesianCalculator::ComputeShortestInterval. This fixes problem reported in ROOT-4619 | 08 February 2019, 14:54:52 UTC |
| f53e622 | Kim Albertsson | 08 February 2019, 14:14:07 UTC | [TMVA] Fix GUI can't find all trained BDT's for plots (#3253) * TMVA-GUI: fix group folder name for methods When writing trained methods to the output file, there was a typo in the target directory. This lead to the results of a method of type "BDT" and name "MyBDT" being placed (incorrectly) in a folder named "Method_MyBDT/MyBDT". This patch places the data correctly in "Method_BDT/BDT". This is important since e.g. the GUI relies on all BDT's being placed in the "MethodBDT" folder. Squashed commits (slight refactoring and fixups): * TMVA: MethodBase -- fix indentation isses * TMVA: MethodBase -- refactor datasetName * TMVA: fix line wrapping * TMVA: refactor for clarity and better naming * clang-format * fix typo in type * `fFactoryBaseDir` is _not_ a field of MethodBase * Fix directory name in envelope test | 08 February 2019, 14:14:07 UTC |
| 3108084 | moneta | 08 February 2019, 08:57:18 UTC | Add a test for the new modified functions TH1::FindFirstBinAbove and FindLastBinAbove | 08 February 2019, 13:50:36 UTC |
| cf6d8ba | moneta | 08 February 2019, 08:56:31 UTC | Re-factor FindFirstBinAbove and FindLastBinAbove by implementing a common function for all histogram dimensions. Add possibility to look in a range, re-implemnting on top of PR #1533 from @ksmith0. | 08 February 2019, 13:50:36 UTC |
| 0eb3194 | Karl Smith | 19 January 2018, 15:32:08 UTC | Added a search range to TH*::FindFirst(Last)Bin. The additional firstBin and lastBin arguments limit the axis being searched for the bin above or threshold. If the firstBin term is less than 1 it is set to 1. If the lastBin term is less than 0 or greater than the number of bins the value is set to the number of bins for that axis. | 08 February 2019, 13:50:36 UTC |
| 5b257e1 | Kim Albertsson | 24 January 2019, 21:29:53 UTC | TMVA: Dataloader includes last cleaning of this batch | 08 February 2019, 13:38:10 UTC |
| d5e72da | Kim Albertsson | 24 January 2019, 21:25:03 UTC | TMVA: Remove more dependencies from DataLoader | 08 February 2019, 13:38:10 UTC |
| 0c6f2ab | Kim Albertsson | 24 January 2019, 21:23:04 UTC | TMVA: Clean up dependencies for DataLoader a bit | 08 February 2019, 13:38:10 UTC |
| c56cad4 | Kim Albertsson | 24 January 2019, 21:05:21 UTC | TMVA: Remove Envelope from being friends with DataLoader | 08 February 2019, 13:38:10 UTC |
| d3c2d46 | Kim Albertsson | 24 January 2019, 20:55:59 UTC | TMVA: Remove Factory as a friend of DataLoader | 08 February 2019, 13:38:10 UTC |
| 4ab83b9 | Paul Seyfert | 04 February 2019, 14:52:34 UTC | adding linebreak in codegen | 08 February 2019, 13:35:22 UTC |
| fbfeef2 | Lars Beemster | 21 January 2019, 17:15:54 UTC | Fix GoesRight() and some Double_t -> double in generated code | 08 February 2019, 13:32:34 UTC |
| 20f4d31 | Lars Beemster | 22 November 2018, 20:20:26 UTC | processed cosmetic comments on PR, use return by value | 08 February 2019, 13:32:34 UTC |
| 2244100 | Lars Beemster | 22 November 2018, 15:47:02 UTC | Implement standalone class generation for multiclass BDTG | 08 February 2019, 13:32:34 UTC |
| bbc7535 | Kim Albertsson | 14 January 2019, 14:49:12 UTC | [TMVA] Users guide -- update instructions for randomised trees | 08 February 2019, 12:59:02 UTC |
| 87b7536 | Axel Naumann | 07 February 2019, 09:37:49 UTC | [core] Check TString length parameters (ROOT-5012). | 07 February 2019, 20:25:18 UTC |
| 4f4ab67 | Bertrand Bellenot | 07 February 2019, 09:02:03 UTC | Get rid of TGWin32InterpreterProxy (was commented out anyway) | 07 February 2019, 14:39:11 UTC |
| fa8d620 | Henry Fredrick Schreiner | 05 February 2019, 13:41:51 UTC | Adding directories needed to use LibXML | 07 February 2019, 13:21:35 UTC |
| dba4356 | Guilherme Amadio | 01 February 2019, 12:28:11 UTC | test-periodic: build with same C++ compiler as ROOT If not built with the same compiler, CMAKE_CXX_FLAGS may contain flags not known by the default compiler, which will make the test fail. In particular using Ninja + Clang on Linux makes this test fail with the error below without this patch. c++: error: unrecognized command line option ‘-fcolor-diagnostics’ ninja: build stopped: subcommand failed. | 07 February 2019, 13:16:13 UTC |
| d1d08b0 | Stephan Hageboeck | 07 February 2019, 10:21:57 UTC | Fix clang-tidy script for CI. clang-tidy would run the wrong checks/no checks at all because of a missing comma. | 07 February 2019, 10:21:57 UTC |
| 238ff7c | Guilherme Amadio | 06 February 2019, 16:40:40 UTC | io: remove dead branch from read-only TMemFile constructor ParseOption("READ") returns EMode::kRead, so we know that the if will never execute, and that we do not need to create the file, since we are reusing external data. Nevertheless, make the file a zombie if the input buffer is empty, since a read-only empty file is otherwise useless. | 07 February 2019, 08:58:13 UTC |
| 3c50a65 | Guilherme Amadio | 06 February 2019, 16:14:51 UTC | io: use TBufferFile with same size as TBufferMergerFile Using nbytes we get a buffer only as big as last thing written with TMemFile::Write(), which causes many memory allocations to expand the TBufferFile to the same size as the TBufferMergerFile. Related JIRA issue: ROOT-9133. | 07 February 2019, 08:58:13 UTC |
| 2b54b2e | Guilherme Amadio | 06 February 2019, 10:40:09 UTC | io: use new TMemFile constructor in TBufferMerger::Merge() This avoids copying data from the buffer into the TMemFile when merging. Related JIRA issue: ROOT-9133. | 07 February 2019, 08:58:13 UTC |
| 908b6f5 | Guilherme Amadio | 06 February 2019, 10:13:56 UTC | io: add constructor to create a read-only TMemFile from TBufferFile TMemFiles can copy themselves into a TBuffer already, so this constructor is more convenient to create a TMemFile with data that was written out by another TMemFile and copied into the TBuffer. Related JIRA issue: ROOT-9133. | 07 February 2019, 08:58:13 UTC |
| e132d71 | Guilherme Amadio | 06 February 2019, 10:10:04 UTC | io: add GetMergeOptions()/SetMergeOptions() to TBufferMerger This is to allow the user to set the merging options to avoid recompression of input data into the output file when the compression algorithm or level are different. ROOT-9779 #resolve | 07 February 2019, 08:58:12 UTC |
| 36b569b | Guilherme Amadio | 06 February 2019, 09:28:21 UTC | io: small fix for TBufferMerger doxygen docs | 07 February 2019, 08:58:12 UTC |
| c7258c9 | Stephan Hageboeck | 07 February 2019, 08:01:19 UTC | [RF][ROOT-9816] Replace core collections of RooFit with std::vector * Implement a RooSTLRefCountList as replacement for RooRefCountList. * Replace performance-critical instances of old ref count list, in particular in RooAbsArg. * Improvement of documentation, clang-tidy. * Remove more uses of legacy iterators over RooAbsCollection, in particular members of RooAbsDataStore, RooRealIntegral, RooSetProxy, RooRealIntegral. | 07 February 2019, 08:01:19 UTC |
| a5e0f39 | Danilo Piparo | 05 February 2019, 16:26:47 UTC | [Core] Add emptypcm argument to RIO, Core and Thread dictionaries this will allow cling to just get the information from the pch without triggering autoparsing of the full payload. | 06 February 2019, 18:49:47 UTC |
| 1d84c81 | Stephan Hageboeck | 06 February 2019, 16:43:04 UTC | [RF][ROOT-9816] Speed up RooAbsCollection * The default collection in RooFit, the RooLinkedList, has been replaced by a std::vector to benefit from contiguous memory. * Replace often-used iterators in RooFit by std::vector iterators and range-based for loops. * Provide legacy iterators with old interface to not break existing code. * Flag legacy iterators with R__SUGGEST_ALTERNATIVE. * Update doxygen and release notes. | 06 February 2019, 16:43:04 UTC |
| 6631c1b | Stephan Hageboeck | 06 February 2019, 16:07:07 UTC | [RF] Silence importing datasets to RooWorkspace. | 06 February 2019, 16:07:07 UTC |
| a52a61a | Enrico Guiraud | 01 February 2019, 12:15:08 UTC | [TBufferMerger] Initialize TBufferFile with the right size in Write | 06 February 2019, 13:36:34 UTC |
| 3e6c3d0 | Stephan Hageboeck | 06 February 2019, 08:21:30 UTC | [RF] Silence RooWorkspace unit test. The unit test for RooWorkspace tests failures when importing models into the workspace. Since RooFit will print `ERROR` messages, it looks like there is a problem in the test. The error messages are now silenced. | 06 February 2019, 08:21:30 UTC |
| 2b36b53 | Olivier Couet | 05 February 2019, 16:52:07 UTC | SetTitle now handles properly the title in the form `stringt;stringx;stringy;stringz` This fixes the issue ROOT-9767 | 05 February 2019, 16:52:22 UTC |
| bc35b1c | Olivier Couet | 05 February 2019, 16:14:26 UTC | - The saved TGraph2D should be access via its pointer to call SetDirectory. | 05 February 2019, 16:52:22 UTC |
| d7dcdc0 | Lorenzo Moneta | 05 February 2019, 15:04:42 UTC | Update documentation for TRandom::Integer as suggested in ROOT-9928 | 05 February 2019, 15:04:55 UTC |
| b8c925b | Danilo Piparo | 04 February 2019, 14:27:28 UTC | [Core] Replace with TClass::GetClass<T>() other invocations of GetClass to show good practices or to lower contention on the global lock. | 05 February 2019, 10:55:12 UTC |
| be82fbc | Danilo Piparo | 04 February 2019, 14:24:33 UTC | [Core] Express {TClass,ROOT}::GetClass methods in terms of GetClass<T>() whenever possible, e.g. with signatures accepting T* and alike. | 05 February 2019, 10:55:12 UTC |
| 0f16f4f | Danilo Piparo | 05 February 2019, 10:54:10 UTC | [DF][NFC] Add jitted filter usage example. | 05 February 2019, 10:54:32 UTC |
| 0e59338 | Olivier Couet | 05 February 2019, 10:31:11 UTC | Reformat the doc to fix ROOT-9923 | 05 February 2019, 10:31:21 UTC |
| b669d9e | Oksana Shadura | 29 January 2019, 19:42:39 UTC | [ROOT-9924] Fixing compile time installation of clad in CMAKE_INSTALL_PATH [ROOT-9924] Installing headers for clad | 05 February 2019, 10:27:36 UTC |
| ed60562 | Sergey Linev | 29 January 2019, 15:24:51 UTC | meta: TGlobalMappedFunction formatting, use override | 05 February 2019, 07:59:16 UTC |
| 2747f3e | Sergey Linev | 29 January 2019, 12:34:45 UTC | meta: use functor for extracting pointer on TIterpreter::Instance | 05 February 2019, 07:59:16 UTC |
| 4fde772 | Sergey Linev | 29 January 2019, 12:31:39 UTC | meta: fix in extracting pointer on gClientGlobal | 05 February 2019, 07:59:16 UTC |
| 55a6662 | Sergey Linev | 29 January 2019, 12:26:44 UTC | meta: use functor for providing gClient TGClient::Instance() does not provide reference like gFile | 05 February 2019, 07:59:16 UTC |
| bc7179c | Sergey Linev | 29 January 2019, 12:22:01 UTC | meta: provide TGlobalMappedFunction::MakeFunctor with functor This special functor should correctly extract pointer on variable. Not all global ROOT functions can provide reference. | 05 February 2019, 07:59:16 UTC |
| 8e17010 | Sergey Linev | 21 December 2018, 15:24:39 UTC | Return address on global variable in TGlobalMappedFunction This is how TGlobal::GetAddress() is defined. Provide workaround for gROOT. TInterpreter::Instance and TGClient::Instance has wrong signature (returns value, not reference), therefore here is not possible to return real address. Keep old code until signature is changed | 05 February 2019, 07:59:16 UTC |
| 401ab18 | Sergey Linev | 21 December 2018, 10:39:45 UTC | Remove unused include | 05 February 2019, 07:59:16 UTC |
| 5ad3484 | Sergey Linev | 21 December 2018, 10:30:07 UTC | "Hide" functor in TGlobalMappedFunction In templated MakeFunctor method functor is created individually for each global functions kind. Per flag one can add created TGlobalMappedFunction object to globals. | 05 February 2019, 07:59:16 UTC |
| f57dfbb | Sergey Linev | 21 December 2018, 10:10:07 UTC | Provide templated TGlobalMappedFunction::AddFunctor() method It takes global function as templated argument as creates appropriate functor with correct casting | 05 February 2019, 07:59:16 UTC |
| c20b096 | Sergey Linev | 21 December 2018, 09:52:17 UTC | TGlobalMappedFunciton: clang-format and nullptr | 05 February 2019, 07:59:16 UTC |
| 5de461a | Sergey Linev | 21 December 2018, 09:31:36 UTC | globals function: use in all places std::function to extract pointer Previous cast of functions had incompatible types: "void* (*)" instead of "void* &(*)" And casting of function types are always dangerous | 05 February 2019, 07:59:16 UTC |
| 4d478eb | Sergey Linev | 21 December 2018, 09:06:59 UTC | [global functions] use std::function<void*()> to extract pointer | 05 February 2019, 07:59:16 UTC |
| 0d6375b | Bertrand Bellenot | 01 February 2019, 13:31:20 UTC | Use `llvm::raw_string_ostream` instead of `std::stringstream` (thanks Axel) | 05 February 2019, 07:05:58 UTC |
| 74ee4a3 | Bertrand Bellenot | 31 January 2019, 08:58:11 UTC | Windows: Fix formatting of Interpreter::toString(const char* type, void* obj) Make sure hexadecimal pointer values have the correct '0x' prefix (not automatic on Windows) | 05 February 2019, 07:05:58 UTC |
| 50aefd6 | Henry Fredrick Schreiner | 04 February 2019, 10:57:41 UTC | Adding back ROOT/RConfig.h to point to the new location | 05 February 2019, 06:57:56 UTC |
| 19e4890 | Henry Fredrick Schreiner | 01 February 2019, 07:41:47 UTC | Rename based fix for RConfig on case-insensitive systems | 05 February 2019, 06:57:56 UTC |
| 3bc90bc | Oksana Shadura | 04 February 2019, 15:20:07 UTC | Importing 'Imt.ROOT/TTaskGroup.hxx' modules required by TThreadExecutor (#3374) | 04 February 2019, 15:20:07 UTC |
| c316d09 | Axel Naumann | 04 February 2019, 09:04:18 UTC | [davix] Be helpful and state file name in open error. | 04 February 2019, 09:47:31 UTC |
| 2fcb082 | Danilo Piparo | 01 February 2019, 12:14:48 UTC | [IO][ROOT-9926] Add support for data members which are unique_ptrs<const T> and add a unit test for TClassEdit. | 04 February 2019, 09:15:13 UTC |
| 65c740f | Enric Tejedor Saavedra | 01 February 2019, 14:49:18 UTC | [Exp PyROOT] Variables already defined by CPyCppyy module | 04 February 2019, 08:17:25 UTC |
| a6e99e6 | Enric Tejedor Saavedra | 01 February 2019, 14:35:39 UTC | [Exp PyROOT] Expose some functionality from cppyy module Several functions that are present in the ROOT module of the current PyROOT now belong to the CPyCppyy module. This commit exposes some of them in the new (experimental) ROOT module, since they are used in a few tests. A more exhaustive investigation needs to be done to understand what cppyy functionality we would like to expose in ROOT. | 04 February 2019, 08:17:25 UTC |
| a3988db | Matevz Tadel | 15 November 2018, 07:42:03 UTC | Support new Hexagon type in TEveBoxSet + 2 bufixes 1. Add support for hexagont box-type (BHex_t) in TEveBoxSet. See tutorials/eve/boxset.C, new function boxset_hex(). 2. Bugfix in TEveBoxSet -- box set did not work with TEnv setting of `OpenGL.UseDisplayLists 0`. 3. Memory leak fix in TEveGeoShapeProjected -- TBuffer3D member fBuff was not freed in destructor. | 01 February 2019, 19:16:15 UTC |
| c52012c | Enric Tejedor Saavedra | 01 February 2019, 17:43:29 UTC | [Exp PyROOT] Make tests runnable as Python scripts | 01 February 2019, 17:44:56 UTC |
| 16c66bb | Enric Tejedor Saavedra | 01 February 2019, 13:32:06 UTC | [Exp PyROOT] Re-enable tutorial, ROOT facade is there | 01 February 2019, 14:51:44 UTC |
| 1af28be | Stephan Hageboeck | 01 February 2019, 14:38:19 UTC | [RF][ROOT-9518][ROOT-5333] Fix analytical integral of RooPoisson [ROOT-9518] The analytical integral of RooPoisson was interpolating between the integer points of the discrete CDF. Due to this, the CDF, is shifted by 1 such that asking for CDF_Pois(x=0, mu=1) yields 0. Further, due to numerical precision problems, the integral can be off by 1.E-2 depending on the range. The CDF obtained with these changes is now discrete. [ROOT-5333] Giving a range with an upper limit larger than max int will overflow an integer and yield NaN. Now, max int will be used as end of the integration range, yielding an integral of 1 in such cases. For an unlimited range, the integral will always be 1. Unit tests (testRooPoisson) have been added to test the integral of RooPoisson against publicly available tables, as well as ROOT::Math::poisson_cdf. The stressRooStats reference file needed to be updated because the fixed implementation will yield different values for tests where a product of two Poisson distributions is evaluated. | 01 February 2019, 14:38:19 UTC |
| 1a99bf2 | Danilo Piparo | 01 February 2019, 11:07:40 UTC | [DF][ROOT-9933] Add GetNSlots method to RInterface and a test. | 01 February 2019, 13:57:16 UTC |
| c4b95a7 | Danilo Piparo | 01 February 2019, 10:46:16 UTC | [DF][ROOT-9931] Properly check available columns when booking Histo1D and Graph | 01 February 2019, 13:57:16 UTC |
| 893cfbe | Stephan Hageboeck | 01 February 2019, 12:32:50 UTC | [RF] Fix spurious binning warning in RooDataHist RooDataHist warns that it has to readjust the binning of a variable although no such adjustment is happening. Fixes: ROOT-8163 | 01 February 2019, 12:32:50 UTC |
| 78019d9 | Stephan Hageboeck | 01 February 2019, 09:38:00 UTC | [RF] Remove fixed-length char buffers from RooWorkspace The RooWorkspace used 'char buf[xx]' in various places to process user strings. Supplying a long string to e.g. import object into a workspace would lead to the rest being ignored silently. Fixes: ROOT-9486 | 01 February 2019, 09:38:00 UTC |
| c27fcf1 | Axel Naumann | 30 January 2019, 08:51:35 UTC | [v7hist] Make fCanGrow non-const, to enable default assignment op: Works around In file included from include/ROOT/RHist.hxx:20: include/ROOT/RAxis.hxx:57:15: warning: explicitly defaulted copy assignment operator is implicitly deleted [-Wdefaulted-function-deleted] RAxisBase &operator=(const RAxisBase &) = default; ^ include/ROOT/RAxis.hxx:250:15: note: copy assignment operator of 'RAxisBase' is implicitly deleted because field 'fCanGrow' is of const-qualified type 'const bool' const bool fCanGrow = false; ///< Whether this axis can grow (and thus has no overflow bins). ^ | 01 February 2019, 09:19:56 UTC |
| 198639c | BerserkerTroll | 17 March 2018, 19:13:43 UTC | Force the caller to use std::move to steal internals | 01 February 2019, 09:19:56 UTC |
| af1ad9a | Sergey Linev | 31 January 2019, 10:27:37 UTC | webgui: allow --web also when ROOTWebDisplay library already load | 01 February 2019, 07:15:53 UTC |
| e2917fa | Danilo Piparo | 31 January 2019, 10:00:18 UTC | [PyMVA][ROOT-9677] Acquire Gil where needed | 31 January 2019, 20:27:12 UTC |
| 4cfd161 | Enrico Guiraud | 31 January 2019, 10:31:24 UTC | [DF][NFC] Minor fixes to OnPartialResult doc | 31 January 2019, 14:11:56 UTC |
| e9fe075 | Enric Tejedor Saavedra | 30 January 2019, 17:00:00 UTC | [Exp PyROOT] Configure preliminar module facade for ROOT This commit puts in place a facade for ROOT to become a Python module, in a similar way it is done for the current PyROOT. For now this is just a simple skeleton that redirects lookups to cppyy's global namespace, and more investigation on the logic implemented by the facade of the current PyROOT is necessary. | 31 January 2019, 09:47:05 UTC |
| 73ffb0d | Stephan Hageboeck | 31 January 2019, 09:22:27 UTC | [RF][ROOT-9777] Replay "Fix cloning workspaces" - When RooWorkspaces that contain a ModelConfig are cloned, the ModelConfig did not get updated. It would still point to the old workspace. - Add unit test. | 31 January 2019, 09:22:27 UTC |
| 9ede664 | Bertrand Bellenot | 31 January 2019, 09:05:27 UTC | Windows: Add missing cling exports | 31 January 2019, 09:05:27 UTC |
| bdfd174 | Sergey Linev | 30 January 2019, 16:54:39 UTC | webgui: check ROOTWebDisplay library when --web option specified When starting ROOT with --web option, check if necessary libraries are build. Only in that case change Gui.Factory and call gROOT->SetWebDisplay. Fixes https://sft.its.cern.ch/jira/browse/ROOT-9927 | 31 January 2019, 08:54:26 UTC |
| 0d22cbc | Sergey Linev | 30 January 2019, 16:23:08 UTC | Silent gcc8 warning when specify callback function for gluTessCallback Unfortunately, gluTessCallback uses same function pointer type for absolutely different functions kinds. Only inside gluTessCallback proper casting done. Therefore the only solution - suppress warings | 31 January 2019, 08:19:59 UTC |
| 154b2e1 | Danilo Piparo | 30 January 2019, 21:11:24 UTC | [DF][ROOT-9929] Extract branches names which are nested from jitted strings properly, for example treating cases such as 'branch0' and 'branch01'. The code of the helper FindUsedColumnNames has been simplified. A test has been added. | 31 January 2019, 07:34:15 UTC |
| 3c6cf1e | Yuka Takahashi | 23 January 2019, 09:21:18 UTC | [cxxmodule] Fix a warning in modules build Not expose ROOT/RConfig.h to rootcling, but to add the header to ROOT_Config pcm. | 30 January 2019, 16:11:22 UTC |
| 1f75abe | moneta | 30 January 2019, 11:39:47 UTC | Use for bin edge comparisons an epsilon abs value that is computed as a fraction of bin width. This would give a more robust comparison test. This should fix what has been shown in ROOT-7752 | 30 January 2019, 15:04:33 UTC |
| fb1357c | Sergey Linev | 30 January 2019, 11:19:22 UTC | geom: create TGeoCone with name in TGeoBuilder Looks like only TGeoCone was ignoring name. Fixes error described in https://root-forum.cern.ch/t/32493 | 30 January 2019, 15:03:55 UTC |
| 247ecf1 | Stephan Hageboeck | 30 January 2019, 14:57:16 UTC | Add R__SUGGEST_ALTERNATIVE macro to flag legacy functions Functions that are outdated, but will not be removed, can be flagged with `R__SUGGEST_ALTERNATIVE`, a macro expanding to ROOT's `_R__DEPRECATED_LATER`. This can be used to point users to the recommended interfaces. Flagging a function like this: TIterator* createIterator() const R__SUGGEST_ALTERNATIVE("begin(), end() and range-based for loops are faster.") { ... } and defining `R__SUGGEST_NEW_INTERFACE` in a translation unit will trigger a warning such as: RooChebychev.cxx:66:34: warning: 'createIterator' is deprecated: There is a superior alternative: begin(), end() and range-based for loops are faster. [-Wdeprecated-declarations] TIterator* coefIter = coefList.createIterator(); IMPORTANT: Theoretically, the macro could be used for classes: Don't! This will trigger warnings for all uses of the class in the interfaces of ROOT. | 30 January 2019, 14:57:16 UTC |
| b7e6379 | Danilo Piparo | 30 January 2019, 12:46:15 UTC | [Graphics][NFC] Clarify the options of TROOT::SetWebDisplay | 30 January 2019, 12:46:15 UTC |
| 68635da | Stephan Hageboeck | 30 January 2019, 12:14:15 UTC | [RF] Reactivate copy assignment for HistFactory classes * Make histfactory Samples copy-assignable. Rootbench depends on copy assignment for the Sample class. After implementing move semantics to speed up hist2workspace, the copy assignment had been implicitly deleted. This was fixed, and a gtest has been added. * Further, HistRef, a member of Sample, is now movable, and the histograms given back by Sample are now const TH1* instead of TH1*. | 30 January 2019, 12:14:15 UTC |
| 77fcbbb | Stephan Hageboeck | 30 January 2019, 12:10:04 UTC | [RF] Improve RooFit tutorials * Temporaries in python: In the RooFit python tutorials, C++ objects save pointers to objects held by python. When these are temporaries, this leads to crashes. Naming the temporaries works around this problem. * rf204a: Add a new tutorial that demonstrates using extended pdfs when fitting in multiple ranges. | 30 January 2019, 12:10:04 UTC |
| b3e0431 | Yuka Takahashi | 29 January 2019, 15:35:31 UTC | [cxxmodule] Infer proper modulename from library Before RegisterModule was assuming that "modulename" starts from "lib", which is not the case for CMSSW where the "modulename" is "SimDataFormatsValidationFormats_xr" for example. | 30 January 2019, 07:28:50 UTC |
| d7f10a9 | Stephan Hageboeck | 29 January 2019, 18:15:50 UTC | [RF] Speed up hist2workspace hist2workspace reads histograms from multiple files, which often contain numerous folders with a lot of histograms. It therefore needlessly suffers from TList::RecursiveRemove slowdowns, although almost all histograms and files are read only. Implementing the memory management for histograms without ROOT's TList, and implementing move semantics with better const correctness leads to a speed up of a factor 10 for a large ATLAS-style Higgs workspace with 28 channels and about 150 systematic uncertainties. | 29 January 2019, 18:15:50 UTC |
| a682c54 | Axel Naumann | 29 January 2019, 13:57:59 UTC | [rint] 2019! Thanks, Stephan. | 29 January 2019, 13:57:59 UTC |
| fc1a587 | Stephan Hageboeck | 29 January 2019, 10:13:54 UTC | RooRealSumPdf | 29 January 2019, 13:51:15 UTC |
| 538e7f6 | Stephan Hageboeck | 29 January 2019, 10:09:36 UTC | RooAbsReal getVal(). | 29 January 2019, 13:51:15 UTC |
| 3452e61 | Stephan Hageboeck | 29 January 2019, 10:09:17 UTC | HistFactory. | 29 January 2019, 13:51:15 UTC |
| c5f214c | Stephan Hageboeck | 14 January 2019, 10:40:29 UTC | [RF] Improve docs of RooPolynomial. | 29 January 2019, 13:51:15 UTC |
| cc4eb63 | Bertrand Bellenot | 22 January 2019, 16:34:25 UTC | Only use the #pragmas if not using Visual Studio | 29 January 2019, 13:44:49 UTC |
| 39ecc32 | Bertrand Bellenot | 22 January 2019, 10:41:46 UTC | Fix "unknown pragma" compiler warning on Windows | 29 January 2019, 13:44:49 UTC |
| f280e2b | Bertrand Bellenot | 22 January 2019, 16:31:41 UTC | Only use the #pragmas if not using Visual Studio | 29 January 2019, 13:44:06 UTC |
