| 75c0d62 | anerokhi | 01 May 2019, 21:21:19 UTC | Fix handling of ':' character in a TDirectory name | 24 October 2019, 15:57:23 UTC |
| 04be753 | Philippe Canal | 24 October 2019, 14:20:57 UTC | Update release notes (hadd speed) | 24 October 2019, 14:21:07 UTC |
| c587105 | Philippe Canal | 15 October 2019, 00:25:37 UTC | [hadd] Delay reset of kMustCleanup to avoid recording the change | 24 October 2019, 14:21:07 UTC |
| 89a5476 | Philippe Canal | 09 October 2019, 17:54:36 UTC | [hadd] Delete the directory from the input file as soon as done with it. So far only the directory for the _first_ input file were delete immmediately, all the other were kept in memory until the files were close. This reduces memory consumption and speedup file closing (tear down) speed. | 24 October 2019, 14:21:07 UTC |
| 2fc42fe | Philippe Canal | 09 October 2019, 17:52:27 UTC | [hadd] Mark all read objects as do-not-cleanup | 24 October 2019, 14:21:07 UTC |
| e6ae50a | Sergey Linev | 15 October 2019, 16:16:37 UTC | TMessage: do not access not-allocated buffer ROOT-10340 When wrong-formatted TMessage is received, length of buffer may be 0. As result, buffer may not be allocated at all. One should avoid that in such case memory is accessed | 15 October 2019, 17:49:44 UTC |
| 3a63bc9 | Philippe Canal | 24 September 2019, 21:36:19 UTC | Fix handling of inline namespace std::__cxx11 in TClassEdit::GetSplit This fixes ROOT-8396 The issues was the normalization of "std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > >" into vector<string> due to the inline namespace (__cxx11). Prior to this commit, it would work but only if the name for normalized twice. This because the order of operation was 1. check if the name is std::basic_string or basic_string 2. do more stuff 3. remove std and std::[any-inlined-namespace-name] now we extended 1 to also search for std::[any-inlined-namespace-name]::basic_string | 25 September 2019, 11:47:22 UTC |
| bb9e08e | Philippe Canal | 03 April 2018, 17:09:08 UTC | Must set fMaxClusterRange (transient) in TTree::Streamer. This lead to corrupted cluster range tables (and thus failure in dataframe_snapshot) when importing the cluster table into a TTree that was just read from file and has a cluster table. | 30 January 2019, 01:19:28 UTC |
| f55005c | Philippe Canal | 29 March 2018, 17:24:34 UTC | Add basketsInfo option to TBranch::Print | 30 January 2019, 01:19:28 UTC |
| 510015d | Philippe Canal | 23 March 2018, 22:53:57 UTC | Insure TTreeCloner keep the AutoFlush history. | 30 January 2019, 01:19:28 UTC |
| 63fb599 | Philippe Canal | 23 March 2018, 22:52:12 UTC | Do not keep AutoFlush history when GetEntries() is 0 | 30 January 2019, 01:19:28 UTC |
| 6b2664e | Guilherme Amadio | 16 April 2018, 08:43:20 UTC | Update builtin version of openssl to 1.0.2o | 16 April 2018, 08:43:51 UTC |
| a992483 | Philippe Canal | 29 March 2018, 19:03:13 UTC | gCoreMutex is not yet available | 29 March 2018, 19:03:13 UTC |
| b036712 | Philippe Canal | 22 March 2018, 15:51:27 UTC | Prevent wild over-read when reading badly clustered file. In some cases where the memory available to the TTreeCache can not fit the entirety of (potentially odd-shaped) cluster, the TTreeCache could end up with a situation where (starting at some point) it is invalidated at each entry *and* the basket from the previous cluster boundary up to some entry number less than the current entry are loaded over and over again (until the entry number reaches the next cluster boundary). In a case encountered by CMS, a TTree has a recorded cluster size of 165 events. However the TTree appears to not have been clustered; most basket size are the same and seem to be flushed at 139 events interval. A few branches are flushed less frequently, for example EventAuxiliary is flushed every 459 entries. As a consequence when entry 417 is requested, the cluster boundary is calculated as 330 and the first end as 495. Then all basket containing entries in that range are loaded, including EventAuxiliary's basket containing entries 459 to 917. Then 917 is the highest available (partial) entry available. Thus the next time the TTreeCache is invalidated is when entry 918 is requested. As that point the cluster boundary is calculated as 845 and the first end as 495, however (with the small enough TTreeCache size), the cache is full as soon as the basket for EventAuxiliary is added. This resulted in a 'valid' range for this cache of 845 to 917 and lead to the reading all the baskets added so far ... and thus eventhough none of them would be used for reading and they already had been read once in memory. Then for entry 919, the exact same behavior repeated. The solution consist on marking the 'minimum' valid range as being 'at least' one cluster wide. Thus in the example above for entry 918 through 989, **only** the missing baskets are read individually resulting in 'degraded' performance but **not** a wild over-read. A future enhancement would be, in those cases, restart the next buffer from the last valid entry + 1 rather than the previous event boundary. This would result in a 'slightly' over-read (some baskets might be read twice) rather than the wild useless over-read we had. | 29 March 2018, 18:49:22 UTC |
| 8622a77 | Philippe Canal | 22 March 2018, 15:40:47 UTC | Actually enable TTree::Print("clusters"). It was unusable due to a typo | 29 March 2018, 18:49:21 UTC |
| 36d0b0a | Philippe Canal | 29 March 2018, 02:42:14 UTC | Simplify and fix thread-safetly in TFormLeafInfoMethod::ReturnTClass | 29 March 2018, 15:21:07 UTC |
| a8b4eca | Philippe Canal | 28 March 2018, 20:03:08 UTC | Restore TTree::Draw ability to use function return a reference to object. For example std::vector::at | 29 March 2018, 15:21:05 UTC |
| e12a366 | Pere Mato | 11 January 2018, 13:32:31 UTC | Remove unwanted errors when looking for R modules | 11 January 2018, 15:08:58 UTC |
| 1ff2f92 | Pere Mato | 11 January 2018, 11:18:18 UTC | Define PYTHIA8DATA if not defined | 11 January 2018, 15:08:01 UTC |
| fb34c6b | Pere Mato | 10 January 2018, 13:30:52 UTC | Fix for ROOT-9180 - If R components Rcpp and RInside are missing the 'r' option should be disabled | 11 January 2018, 08:43:00 UTC |
| 9c84903 | Pere Mato | 08 January 2018, 14:02:48 UTC | Fix for ROOT-7650 - Complete the 'hdfs' option with CMake | 10 January 2018, 07:36:30 UTC |
| 4373d0d | Bastian Beischer | 01 October 2017, 07:13:18 UTC | TTree::Draw(): Respect binning when drawing TProfile2D with 'col'. When drawing a TProfile2D with TTree::Draw() the histogram Y-axis binning was not set correctly in case the options 'col' and 'prof' where used at the same time in the TTree::Draw() option argument. For example: tree->Draw("var:y:x>>h(100,-1,1,200,-2,2)", "", "colzprof") produced a TProfile2D with 100 bins from -1 to 1 in X, but only 40 bins from -2 to 2 in Y. This fixes ROOT-9034. | 24 November 2017, 08:22:07 UTC |
| 3784039 | Mario Lassnig | 22 November 2017, 11:43:37 UTC | TDavixFile: add support for cross-protocol metalinks Original patch by Georgios Bitzes, manual rebase onto master. (cherry picked from commit 0ddce09a6fd7a25c19de07f1c5a5fe7dd8fba7dc) | 23 November 2017, 15:32:25 UTC |
| 7ad9b54 | Guilherme Amadio | 16 November 2017, 10:28:36 UTC | Set CMake policy CMP0068 to NEW for LLVM This change is to avoid warnings from CMake 3.9.3 and newer on macOS. See cmake --help-policy CMP0068 for more information. Fixes ROOT-9031. | 16 November 2017, 10:31:49 UTC |
| 9a01cb7 | Oksana Shadura | 03 October 2017, 16:19:51 UTC | Fixing Ninja error about multiple rules generate liblz4.a | 03 October 2017, 18:11:35 UTC |
| 392ddcf | Brian Bockelman | 13 July 2017, 14:36:56 UTC | Pass ROOT CMake variables to LZ4 build. This ensures that the LZ4 internal build is built with the same compiler as the rest of ROOT. | 03 October 2017, 18:11:35 UTC |
| c9f805c | Guilherme Amadio | 07 September 2017, 09:53:56 UTC | Update CMake policy settings to CMake version 3.4.3 The command cmake_minimum_required(VERSION 3.4.3) implicitly calls cmake_policy(VERSION 3.4.3), which means that policies created in CMake versions prior to 3.4.3 will use the NEW setting by default, and newer policies will generate warnings. After this change, only CMP0068 (CMake 3.9) still needs to be set explicitly as using NEW behavior. 1. https://cmake.org/cmake/help/latest/command/cmake_policy.html | 28 September 2017, 19:03:06 UTC |
| d7a2be0 | Guilherme Amadio | 28 September 2017, 19:00:16 UTC | Fix CMake warnings regarding policy CMP0022 | 28 September 2017, 19:03:05 UTC |
| ea86ad8 | Danilo Piparo | 20 April 2017, 06:53:13 UTC | Remove unused variable causing a compilation warning | 28 September 2017, 19:03:05 UTC |
| c99deb4 | Oksana Shadura | 27 September 2017, 13:12:04 UTC | Fixing cherry-picking for LZ4 support for makefile ROOT v6-08-00 (classic build) | 28 September 2017, 04:21:25 UTC |
| 93b2ef0 | Oksana Shadura | 04 September 2017, 11:58:14 UTC | Extending FindLZ4.cmake with xxhash includes/library needed for LZ4 checksum functionality | 26 September 2017, 20:00:49 UTC |
| 83bdcae | Oksana Shadura | 10 August 2017, 17:19:55 UTC | Adding support of LZ4 in hadd | 19 September 2017, 20:58:59 UTC |
| 318ea9c | Oksana Shadura | 04 September 2017, 15:09:58 UTC | Adding LZ4 builtin definitions to separate builds with LZ4 builtin and external xxhash | 19 September 2017, 20:58:59 UTC |
| fda9ff6 | Oksana Shadura | 04 September 2017, 17:15:16 UTC | Adding in addition to LZ4-builtin hxxhash, support to pure xxhash API for builtin_lzma=OFF case | 19 September 2017, 20:58:59 UTC |
| c047cb4 | Guilherme Amadio | 05 September 2017, 08:19:41 UTC | Fix classic build when using builtin LZ4 | 19 September 2017, 20:58:59 UTC |
| 59d3628 | Oksana Shadura | 02 September 2017, 14:08:27 UTC | Fixing includes for classic build LZ4 after adding new checksum feature | 19 September 2017, 20:58:59 UTC |
| ff11980 | Brian Bockelman | 30 May 2017, 03:45:56 UTC | Add LZ4 support to classic configure scripts. | 19 September 2017, 20:58:59 UTC |
| 06c6ace | Brian Bockelman | 30 August 2017, 14:10:42 UTC | Formatting fixes for clang-format. | 19 September 2017, 20:58:59 UTC |
| 4870724 | Brian Bockelman | 30 August 2017, 05:15:16 UTC | Add checksum to the LZ4 compressed buffer format. Simply reuses the XXHASH implementation from the LZ4 library (also used in the LZ4 frame format). | 19 September 2017, 20:58:59 UTC |
| 60161b2 | Oksana Shadura | 02 July 2017, 11:22:43 UTC | Fixes in CMakefiles after merging | 19 September 2017, 20:58:59 UTC |
| 84deee7 | Oksana Shadura | 30 June 2017, 13:36:41 UTC | Fixing merge conflicts | 19 September 2017, 20:58:59 UTC |
| 6a6eef5 | Danilo Piparo | 30 May 2017, 11:06:18 UTC | Treat lz4 as all other built-ins | 19 September 2017, 20:58:59 UTC |
| bc8d00c | Brian Bockelman | 25 May 2017, 22:39:31 UTC | Whitespace-only changes for clang-tidy. | 19 September 2017, 20:58:59 UTC |
| 6df32e9 | Brian Bockelman | 25 May 2017, 22:37:19 UTC | Do not use Unix-specific flag on Windows. | 19 September 2017, 20:58:59 UTC |
| ece0d99 | Brian Bockelman | 25 May 2017, 22:36:48 UTC | Make use of TTreePlayer optional. | 19 September 2017, 20:58:59 UTC |
| b0babd2 | Brian Bockelman | 25 May 2017, 22:36:14 UTC | Rename xz to lzma to be self-consistent. | 19 September 2017, 20:58:59 UTC |
| 8c737e4 | Brian Bockelman | 25 May 2017, 22:12:12 UTC | Refactor common logic into helper functions. | 19 September 2017, 20:58:59 UTC |
| 18b3cdd | Brian Bockelman | 25 May 2017, 22:12:05 UTC | Utilize constant, not hardcoded integer. Add LZ4 support to core library. | 19 September 2017, 20:58:59 UTC |
| e0770a3 | Brian Bockelman | 25 May 2017, 20:41:21 UTC | Touch-up whitespace per clang-tidy. Update copyright date on files new to this branch. Add comment explaining logic in CMakeFile.txt Mostly whitespaces; of note, the re-definition of R__likely/unlikely has been moved back into RConfig.h. | 19 September 2017, 20:58:59 UTC |
| ad919d7 | Brian Bockelman | 23 May 2017, 14:56:19 UTC | Hardcode LZ4 tarball URL to point at GitHub. | 19 September 2017, 20:58:59 UTC |
| 24a388e | Brian Bockelman | 11 March 2015, 12:49:16 UTC | Add support for LZ4 as a compression format. Add lz4 source tarball for built-in version. Tweak Event example to make it easier to compare compression algorithms. Tweak build files for Event example to include TreePlayer (for TTreePerfStats). Until we're sure we need it, no special path for Win32. Add -fPIC flag to compile lz4 | 19 September 2017, 20:58:59 UTC |
| f6d9026 | Philippe Canal | 28 August 2017, 20:18:21 UTC | Remove unnecessary search in GetMissingDictionariesForMembers. If TDataMember has a TDataType then the member has a basic type. TDataType::Class 'just' returns the TClass representing TDataType itself. | 28 August 2017, 20:47:56 UTC |
| 91d5c0a | Philippe Canal | 28 August 2017, 20:16:39 UTC | Actually activate GetMissingDictionariesForBaseClasses. It was inadvertently useless as TBaseClass::Class() returns the TClass describing TBaseClass itself while TBaseClass::GetClassPointer returns the TClass describing the base class. | 28 August 2017, 20:47:54 UTC |
| 9565886 | Philippe Canal | 28 August 2017, 20:15:24 UTC | White space | 28 August 2017, 20:47:51 UTC |
| 6fb5cef | Philippe Canal | 25 August 2017, 18:47:21 UTC | Update release notes | 25 August 2017, 18:47:21 UTC |
| 18d854e | Philippe Canal | 28 April 2017, 17:47:04 UTC | Avoid leaking 'old' TProtoClass | 25 August 2017, 18:45:43 UTC |
| b494f0d | Philippe Canal | 16 August 2017, 18:30:50 UTC | Fix ROOT-8896 (Problem with string_view's operator<<) | 16 August 2017, 18:34:10 UTC |
| 31d3b15 | Philippe Canal | 08 August 2017, 16:46:27 UTC | Fix ROOT-8946 (TViewPubDataMembers finding wrong element). This was due to THashList::GetListForObject returning a list that contains more than just the request objects. | 08 August 2017, 17:31:49 UTC |
| 5d1f785 | Philippe Canal | 22 June 2017, 21:27:16 UTC | Update Release notes | 22 June 2017, 21:27:16 UTC |
| 162c98d | Sandro Wenzel | 14 June 2017, 10:24:02 UTC | Fixing an O(N^2) scaling problem caused by TTree::Draw() * TTreeFormala::GetRealInstance repeatedly went trough a while loop that was trasversed again and again for each new query (with lineary increasing instances). This led to a real problem when drawing on a branch that consistent of many data objects (themselves containing small variable sized containers). * The problem is fixed by caching the state of a previous call. | 22 June 2017, 21:25:47 UTC |
| 42b847b | Mattias Ellert | 12 May 2017, 12:24:52 UTC | Use full path for loading CDLL | 16 June 2017, 15:00:58 UTC |
| e2951b4 | Axel Naumann | 17 May 2017, 10:42:14 UTC | These days, freetype depends on -lz. (cherry picked from commit a41483a1ecd61f77751875ae3a3ad0c4c3636449) | 08 June 2017, 07:54:56 UTC |
| 40389fb | Philippe Canal | 03 June 2017, 20:44:23 UTC | Insure that the TBranch Streamer is used in GetTotalSize. This insures that the already-written basket are not included. | 03 June 2017, 21:06:30 UTC |
| a10090c | Philippe Canal | 03 June 2017, 10:43:10 UTC | In TBranch::Streamer insure that we never steam already basket already written to disk | 03 June 2017, 21:06:30 UTC |
| 5b403bb | Philippe Canal | 12 May 2017, 10:37:36 UTC | Remove unused variable | 03 June 2017, 09:38:28 UTC |
| 79dbf2d | Peter van Gemmeren | 24 February 2017, 22:41:02 UTC | Small change to TTree::OptimizeBaskets to avoid disc reads | 02 June 2017, 20:07:35 UTC |
| fc98e39 | Pere Mato | 31 May 2017, 19:38:48 UTC | ROOT-8811 - python: cannot pickle ROOT.Long() (not found as ROOT.long) | 31 May 2017, 19:43:55 UTC |
| fe7763a | Pere Mato | 31 May 2017, 15:23:50 UTC | Fix for ROOT-8809 - push_back on a vector of pointers may not work in PyROOT | 31 May 2017, 15:45:12 UTC |
| 91b8ef6 | Pere Mato | 30 May 2017, 23:16:10 UTC | Fix for ROOT-8805 - itemsize not set on buffers returned by PyROOT functions | 31 May 2017, 09:20:18 UTC |
| 5c29813 | Philippe Canal | 28 May 2017, 18:57:10 UTC | Fix ROOT-8804 (crash in InitCounter function) The problem was induced by a class name and normalized class name being different but being mistaken for a case of schema evolution | 28 May 2017, 20:00:31 UTC |
| c5d5840 | Philippe Canal | 28 May 2017, 16:58:23 UTC | Be more precise about the type needed by TStreamerElement::Init | 28 May 2017, 20:00:31 UTC |
| 0654dcd | Philippe Canal | 25 May 2017, 17:02:13 UTC | GMDLParser: Parse floating point value followed only by white space with stod. This has the side-effect of greatly reducing (actually eliminating in the case reported at ROOT-ABCD) the number of calls made by the GDMLParser into the interpreter. This is turn help avoid triggering the kernel bug (present in 3.10, fixed in 4.11) It is a kernel bug where memory pages that moved back from the page file *sometimes* get the executable bit (or some other page table information) copied incorrectly and/or not on time. We have verified that the problem is seen for both bare-metal and VM running a 3.10 kernel and is *not* seen in neither bare-metal and VM running a 4.11. Some of the changes in the kernel source code between the two version are explicitly race condition in the page fault/retiring handler that lead to already moved page being written to. | 25 May 2017, 19:54:45 UTC |
| 3a495ad | Axel Naumann | 18 May 2017, 06:46:22 UTC | Use __STRICT_ANSI__ and __FAST_MATH__ as set during compilation of cling. (cherry picked from commit 2da9530fa69e17f0a4d95aecfad55430d244c806) | 19 May 2017, 16:02:57 UTC |
| ff9a7c0 | Philippe Canal | 16 May 2017, 14:42:14 UTC | Address ROOT-8795 by locking the access to fSpecials in TROOT::FindObject. CMS has been seeing a periodic crash involving TH2::DoProjection while another thread happens to be doing a minimization. The incidents can be seen here https://github.com/cms-sw/cmssw/issues/18620 The problem is the following: TH2::DoProjection calls TROOT::FindObject https://root.cern.ch/doc/master/TH2_8cxx_source.html#l02086 TROOT::FindObject reads fSpecials without taking the lock https://root.cern.ch/doc/master/TROOT_8cxx_source.html#l01202 while TMinuit manipulates the list of specials while holding a lock. E.g. https://root.cern.ch/doc/master/TMinuit_8cxx_source.html#l00463 If TH2::DoProjection is running concurrently as TMinuit is begin used, it leads to a data race which causes the crash. | 16 May 2017, 14:52:04 UTC |
| 687d51e | Philippe Canal | 15 May 2017, 20:27:24 UTC | Update release notes | 15 May 2017, 20:27:24 UTC |
| 76a461c | Philippe Canal | 17 April 2017, 17:15:01 UTC | Avoid nullptr dereference | 15 May 2017, 19:40:55 UTC |
| f7fd2e0 | Philippe Canal | 17 April 2017, 16:10:28 UTC | Correct out-of order lock acquisition. Previously we had: TFile::Open TFile::GetType TUnixSystem::IsPathLocal TSystem::FindHelper TPluginHandler::CheckForExecPlugin ** take the TPluginHandler lock ** TPluginHandler::SetupCallEnv TClass::GetClass ** take the ROOT/Interpreter lock ** vs TFile::Open TPluginHandler::ExecPluginImpl ** take the ROOT/Interpreter lock ** TMethodCall::Execute ... interpreter layer TPluginManager::AddHandler ** take the TPluginHandler lock ** | 15 May 2017, 19:24:46 UTC |
| 88b8054 | Philippe Canal | 17 April 2017, 16:09:09 UTC | Set the kNotGlobal bit when removing a TF1 from the global list. This avoids spurrious warning about inconsitent state in the case where multiple thread create (for fitting) multiple function with the same name and remove them 'concurrently' with AddToGlobalList(false). | 15 May 2017, 19:24:42 UTC |
| 50ad534 | Philippe Canal | 13 May 2017, 04:11:16 UTC | Mention TMinuit thread friendliness | 13 May 2017, 04:11:16 UTC |
| c583e7f | Philippe Canal | 11 May 2017, 21:11:49 UTC | Make mnrn15 consistent upon usage by multiple thread. The results are now correct but not reproduceable, the stream of random number in all the thread are intermingled. | 13 May 2017, 02:25:50 UTC |
| 4e4441a | Philippe Canal | 11 May 2017, 20:55:48 UTC | Add missing file static to hide variable | 13 May 2017, 02:25:49 UTC |
| 0c9197d | Philippe Canal | 11 May 2017, 20:51:42 UTC | Remove unnecessary static | 13 May 2017, 02:25:49 UTC |
| b931f41 | Philippe Canal | 11 May 2017, 20:48:02 UTC | Remove unnecessary goto | 13 May 2017, 02:25:48 UTC |
| f6a4e0c | Philippe Canal | 11 May 2017, 20:41:26 UTC | Make const expression constexpr | 13 May 2017, 02:25:48 UTC |
| 5071e51 | Philippe Canal | 11 May 2017, 20:39:21 UTC | Remove unnecesarry static or make them const | 13 May 2017, 02:25:47 UTC |
| 659bf50 | Pere Mato | 11 May 2017, 19:45:28 UTC | Fix for ROOT-8792 - Compilation with builtin XRootD fails with GCC 7 | 11 May 2017, 19:52:50 UTC |
| e3bde5c | Axel Naumann | 04 May 2017, 13:20:15 UTC | Dyn scopes: skip scopes nested in the wrapper (ROOT-8773). (cherry picked from commit ab4b24b439c6aaee8a8bf6b810be38b6069c96a2) | 04 May 2017, 13:21:57 UTC |
| f7b3366 | Enric Tejedor Saavedra | 26 April 2017, 07:49:21 UTC | Process sequentially special branch fBranchRef. As pointed out by Philippe: if requested by the user, a special branch (recorded in TTree::fBranchRef) is created that keeps track of which branch contains objects that are 'referenced' by TRef. Upon reading and then dereferencing of a TRef object, TRef::GetObject is called which turns around and calls TBranchRef::Notify which will call GetEntry on the branch. There might be a race condition in the multi-threaded TTree::GetEntry if TRef::GetObject is called during its execution. This is possible if the user inserts schema-evolution rules that use the TRef. So it might be wise to pre-emptively load the TBranchRef in the sequential section. | 02 May 2017, 14:07:07 UTC |
| 3b5851e | Enric Tejedor Saavedra | 24 April 2017, 12:25:18 UTC | Fix for tutorial-multicore-imt001_parBranchProcessing. In the input file of the aforementioned tutorial, some top-level branches depended on another branch that determined their size for every entry (e.g. the size of an array branch). When processing this kind of data, branches that depend on size (or count) branches will trigger a call to TBranch::GetEntry on the latter to ensure their size is read. The scenario described above made the parallelization of TTree::GetEntry thread-unsafe, since a count branch could be processed at the same time by two tasks: its own task and the task that was processing the branch that needed the size. This has been solved by factoring out the count branches, which are processed first sequentially and independently of the rest of the branches. The latter can be processed safely in parallel afterwards because all the sizes have been read already. | 02 May 2017, 14:06:52 UTC |
| 9d7a01f | Axel Naumann | 28 April 2017, 16:18:11 UTC | Update TCollectionProxy/fEnv when reading a new entry (ROOT-8747). Before, Read() relied on the caller to do the Pop/PushProxy pair, and that rarely happend. (cherry picked from commit f1f19924590f7a180b1b4140cb430686bcb57364) | 02 May 2017, 11:59:53 UTC |
| c7d2631 | Pere Mato | 27 April 2017, 09:03:54 UTC | Added Float16_t as fundamental type for PyROOT. Completes ROOT-8732. | 27 April 2017, 09:08:47 UTC |
| 03d9b19 | Pere Mato | 26 April 2017, 16:55:47 UTC | Adding explicitely the impicit fortran libraries. This fixes the problem observed on MacOS with latest versions of gfortran. | 26 April 2017, 16:57:41 UTC |
| 8715261 | Guilherme Amadio | 19 April 2017, 08:22:10 UTC | [TMVA] Prevent double->float conversions in TestMatrixArithmetic Casting the matrices from double to float create small rounding errors of the order of 1e-06 in the reference matrices themselves, so when they are multiplied the sum over long rows/columns accumulates many rounding errors and make the test fail. Using TMatrixT of float as the reference to begin with eliminates this problem. | 21 April 2017, 08:28:14 UTC |
| 5774ffa | Pere Mato | 20 April 2017, 21:41:46 UTC | Fix for ROOT-8732 (#517) * Fix for ROOT-8732 - PyROOT cannot read/write C++ class fields of Double32_t * Try to please clang-format * Again :-( | 20 April 2017, 21:42:53 UTC |
| 6eb2b67 | Gerardo GANIS | 19 April 2017, 15:31:15 UTC | rootd: do not use popen() to expand path names (#513) Addressed security threat reported by S. Luders. Using the same technology used in TSystem::ExpandFileName (with clang-format processing). | 19 April 2017, 15:32:45 UTC |
| 0a96bfe | Pere Mato | 19 April 2017, 14:31:31 UTC | Fix for ROOT-8752- ROOT_GENERATE_DICTIONARY generates rootmap with wrong libsuffix on MacOS | 19 April 2017, 14:39:10 UTC |
| 1e97147 | Danilo Piparo | 06 April 2017, 12:51:49 UTC | [IO] Fix ROOT-8742 cannot safely run on a tree in different threads even if the trees come from files opened in each thread. The solution consists in protecting the TBranchElement::SetAddress. | 18 April 2017, 05:41:27 UTC |
| 728bf7b | Danilo Piparo | 18 April 2017, 05:39:25 UTC | Update copyright notice | 18 April 2017, 05:40:41 UTC |
| b6e22d2 | Pere Mato | 17 March 2017, 10:38:14 UTC | Fix for the precision on ARM64 | 13 April 2017, 10:26:05 UTC |
| 03a32ab | Axel Naumann | 17 February 2017, 10:16:30 UTC | Attempt to fix roottest_root_math_genvector_coordinates4D on ARM64. | 13 April 2017, 10:26:05 UTC |
