ff27713 | Axel Naumann | 02 March 2017, 13:46:50 UTC | Update release notes for v6.08/06. | 02 March 2017, 13:46:50 UTC |
772ec61 | Pere Mato | 28 February 2017, 20:27:06 UTC | Fix for ROOT-8694 - Segfault when drawing TKDE | 28 February 2017, 21:50:52 UTC |
7c31349 | Jan Musinsky | 27 February 2017, 15:53:11 UTC | base/unix: improved GetHostByName/GetSockName/GetPeerName functions | 28 February 2017, 17:57:18 UTC |
ce4b750 | Lorenzo Moneta | 28 February 2017, 15:28:25 UTC | Apply some fixes in DNN introduces after a commit by Simon the 30/10/2016 ( 45d524bafdc9289a7d80b4bd6bdf8006f4b96931) - regularisation was by default now not enabled in the TNet::Loss functions - dropout probabilities were not corrected in the CPU DNN architecture | 28 February 2017, 15:48:43 UTC |
13faa0a | Lorenzo Moneta | 28 February 2017, 15:26:12 UTC | Fix the reduction functions in DNN by performing a serial reduction to ensure a deterministic result in the sum | 28 February 2017, 15:48:42 UTC |
b638978 | Pere Mato | 28 February 2017, 15:11:35 UTC | Retry once more for Segmentation fault | 28 February 2017, 15:12:18 UTC |
5a9694b | Axel Naumann | 28 February 2017, 08:45:03 UTC | Enumerate only DYLIBS (ROOT-8511). (cherry picked from commit d52cd17aaad4b6d3416e41fe6c127877c8442adc) | 28 February 2017, 08:46:59 UTC |
0b85b79 | Danilo Piparo | 27 February 2017, 17:53:32 UTC | [Doc] Release notes | 27 February 2017, 17:53:32 UTC |
46f3085 | Mattias Ellert | 14 January 2017, 17:44:00 UTC | Don't wait for user interaction in batch mode | 27 February 2017, 10:40:41 UTC |
293564d | Danilo Piparo | 24 February 2017, 10:59:04 UTC | [I/O] Fix for [ROOT-8686] | 27 February 2017, 08:56:12 UTC |
5f6ab71 | Danilo Piparo | 22 February 2017, 16:03:48 UTC | [I/O] Fix ROOT-8595 | 27 February 2017, 08:56:07 UTC |
45043a2 | David Abdurachmanov | 26 October 2016, 13:47:08 UTC | Set code model to Large for PowerPC64 (aka ppc64le) This is needed because TOC and text sections can be more than 2GB apart. LLVM SectionMemoryManager is not aware of the design limits of ppc64le while allocating memory for JIT'ed sections. DSO limit was set to 2GB by design. While running CMSSW ROOT/Cling puts TOC and .text.func ~2.7GB apart in VA space. Usually was triggered by TFormula in CMSSW. IBM has modified global entry function prologue. TOC is now stored in 64-bit value before global entry to allow addressing beyond 2GB. This was merged to Clang months ago, but is only applicable to large code model. Later IBM propagated this further: - binutils patches are here: https://sourceware.org/ml/binutils/2015-11/msg00232.html https://sourceware.org/ml/binutils/2015-11/msg00233.html and will be available with binutils 2.26 - GCC patch is here: https://gcc.gnu.org/ml/gcc-patches/2015-12/msg00355.html and will be available with GCC 6.0. - LLVM patch is here: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20160111/324454.html and will be available with LLVM 3.8. - Kernel patches (module loader etc.) are here: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=a61674bdfc7c2bf909c4010699607b62b69b7bec https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=2e50c4bef77511b42cc226865d6bc568fa7f8769 and will be available with Linux 4.5. Signed-off-by: David Abdurachmanov <David.Abdurachmanov@cern.ch> (cherry picked from commit 7be783b223eb3da7369f19844ef540c3005d6a42) | 27 February 2017, 08:38:25 UTC |
32b8e41 | Philippe Canal | 24 February 2017, 20:38:48 UTC | Update release notes | 24 February 2017, 20:38:48 UTC |
e363cc2 | Philippe Canal | 24 February 2017, 20:37:23 UTC | Remove inadvertently added file | 24 February 2017, 20:37:23 UTC |
baaf8d1 | Philippe Canal | 20 February 2017, 21:37:03 UTC | Remove kBOOL_t and thus remove inconsistencies. In 2004 kBool_t was the same value as kUChar_t (both 11). So when the collection proxy were introduced to distinguished the two, a new one was introduced kBOOL_t (value 21) to be used internally by the collection proxies. However this value is returned by TVirtualCollectionProxy::GetType and lead to incorrect behaviors, for example: std::cerr << TDataType::GetDataType(proxy->GetType())->AsString(proxy->At(i)) << "\n"; as the value 21 is in TDataTypes.h the value of kDataTypeAliasUnsigned_t In 2005 (eab7cae079e), kBool_t was updated in TDataTypes.h (to the value 18) to be distinct from kUChar_t as (at the time) sizeof(bool) on MacOS was 4. However the code in collection proxy was never modernize to leverage this new distinction. Since the value is only recorded in the CollectionProxy objects and those are never persisted, we can safetly switch from kBOOL_t to kBool_t. | 24 February 2017, 19:25:30 UTC |
2caaf15 | Philippe Canal | 24 February 2017, 19:14:06 UTC | Allow the use of 1st gen Collection Proxy for bitset and vector<bool> Because the address of the element of a bitset and vector<bool> can not be taken, we need to use a temporary storage to implement the interface that requires the address of the element. So far this had only been implemented in the specialized collection proxy (TGenBitsetProxy and TGenVectorBoolProxy). Those are obtained only when cloning a generic collection proxy. Consequently using the result TClass::GetCollectionProxy for those two directly did not work in all cases/interfaces. However since TTree is making a copy of the collection proxy, it was correctly in all cases. This patch resolves all the issues seen in ROOT-8574 including - the bitset proxy's iterator did not record it started but still recorded its advance - the genertic proxy At function in case of bitset or vector<bool> was not return the right address We also simplified TGenVectorBoolProxy since we know the actual container type at compilation time. | 24 February 2017, 19:25:30 UTC |
87de6d6 | Olivier Couet | 11 January 2017, 12:20:05 UTC | - In the following example, `TPad::WaitPrimitive` was not stoping the macro execution after each plot : ~~~ {.cpp} { TCanvas c1("c1"); TFile f("hsimple.root"); hpx->Draw(); gPad->WaitPrimitive(); hpxpy->Draw(); gPad->WaitPrimitive(); hprof->Draw(); ~~~ this was reported [here](https://root.cern.ch/phpBB3/viewtopic.php?f=3&t=22957). | 24 February 2017, 17:18:47 UTC |
eba8cf2 | Axel Naumann | 20 February 2017, 10:44:43 UTC | Set error flag if .x not found (ROOT-8579). (cherry picked from commit 8986a630e2e61ad5167a50f65c2d40b45966ae23) | 20 February 2017, 11:02:49 UTC |
1660f1f | Mattias Ellert | 17 February 2017, 15:28:26 UTC | Missing header | 17 February 2017, 16:02:32 UTC |
7d95449 | Lorenzo Moneta | 17 February 2017, 11:08:47 UTC | Apply fix for ROOT-8567 also for Profile histograms. Do nothing in TProfile::LabelDeflate when not needed, i.e. when last bin containing the label is the last bin. | 17 February 2017, 11:10:03 UTC |
beeb8c2 | Pere Mato | 14 February 2017, 09:05:23 UTC | Force the lib location for builtin_gsl (to avoid /lib64 in some systems) | 14 February 2017, 09:17:31 UTC |
185d7b0 | Danilo Piparo | 13 February 2017, 10:45:01 UTC | [Doc] Release Notes | 13 February 2017, 10:45:01 UTC |
d00049e | Danilo Piparo | 13 February 2017, 08:46:13 UTC | [DictGen] More expressive error messages when selecting std::arrays directly | 13 February 2017, 08:46:13 UTC |
84379ff | Danilo Piparo | 10 February 2017, 16:47:11 UTC | [IO] Fix column-wise write of std arrays under certain circumstances, column-wise write of instances of classes with an std array as data member could not be written (error prompted about RealData not found). This was fixed: - TRealData.cxx: TRealData::GetName changed so to obtain the correct name in presence of std::array - TCling.cxx: In TCling::InspectMembers does not descend into std::arrays - TClass.cxx: TBuildRealData detects if the datamember is of type std::array and builds the RealData with the correct name - TProtoClass.cxx: Analogous treatment when creating the RealData for initialising the ProtoRealData | 13 February 2017, 08:44:39 UTC |
a3d8895 | Lorenzo Moneta | 10 February 2017, 15:02:14 UTC | Do nothing in TH1::LabelDeflate when not needed, i.e. when last bin containing the label is the last bin. This fixes ROOT-8567 | 10 February 2017, 15:06:25 UTC |
ccaf737 | Danilo Piparo | 08 February 2017, 23:15:23 UTC | TThreadedObject: avoid useless copies at construction time this optimisation can become relevant when the model is built via a copy constructor. | 10 February 2017, 08:22:10 UTC |
2160809 | Ivana Hrivnacova | 07 February 2017, 16:12:40 UTC | Outline TVirtualMCApplication::Instance() - workaround for ROOT-8539 (cherry picked from commit a35d3aed05dc0b1d879bf5c238ef08e7c563e9fb) | 08 February 2017, 10:59:23 UTC |
e3abd23 | Philippe Canal | 31 January 2017, 21:36:16 UTC | Remove previous (incomplete) solution to fix the unwind of JIT stack trace. This is superseded by the enabling/implementation of a section level allocator | 01 February 2017, 15:49:07 UTC |
a5efcd0 | Philippe Canal | 31 January 2017, 18:23:14 UTC | Implemenation a section level allocator in cling's RTDyldMemoryManager (Azog). This allow for a single allocation at the beginning of the module and insure that all the section of the module with be contiguous in memory. This is a better fix for Fix ROOT-8523 and replace commit 3f7418269714e12d9fc4f17d197fde7199dbae7d (see more detail about the problem it the log of that commit) | 01 February 2017, 15:49:07 UTC |
2f2c2ee | Philippe Canal | 25 January 2017, 21:20:56 UTC | Update release notes | 25 January 2017, 21:20:56 UTC |
3e39108 | Philippe Canal | 25 January 2017, 21:07:36 UTC | Correct cause by multiple dictionary for an STL collection. Previous if the user did (implicitly or explicitly) - load first dictionary for a class (without a rootpcm) - access the TClass for that class - load a second dictionary (with a rootpcm) for the same class - Request the list of data members or the list of base class for that class. then the pointer to either list was a nullptr (rather than a pointer to the proper list as provided by the first dictionary). | 25 January 2017, 21:19:14 UTC |
66dd80f | Philippe Canal | 25 January 2017, 20:24:25 UTC | Update release notes | 25 January 2017, 20:24:25 UTC |
e3c4580 | Philippe Canal | 23 January 2017, 20:46:02 UTC | Give up after too many iteration | 25 January 2017, 20:21:56 UTC |
e78aec7 | Marcel R | 21 January 2017, 15:23:37 UTC | Implement 'add' functions from RooAbsCollection in RooArgSet. | 25 January 2017, 18:49:57 UTC |
76deae2 | Gerardo Ganis | 24 January 2017, 15:29:32 UTC | Update release notes | 24 January 2017, 15:29:32 UTC |
16aff10 | Gerardo Ganis | 24 January 2017, 11:37:52 UTC | io/io: fix for ROOT-8516 (corruption in file writing) When the write-cache required flushing to file and the length of the buffer to be written exceed the size of the cache buffer, requiring direct-write-to-file, the offset (seek position) was incorrectly left to the one of the last basket in the flushed cache. | 24 January 2017, 15:19:16 UTC |
98a1ad7 | Philippe Canal | 20 January 2017, 21:25:34 UTC | Update Release notes | 20 January 2017, 21:25:34 UTC |
3478512 | Philippe Canal | 20 January 2017, 18:59:38 UTC | Correct TClass search in TBranchElement::Unroll. Avoid duplicated effort (redo search from class name to TClass for StreamerElement) in particular since the duplication was only partial and contain deficiencies. In particular TBranchElement::Unroll did not support const data member. This was seen by CMS in a case of a class A containing an object of class B containing an ojbect of class C which has const data member. TBranchElement::Unroll segfault in this case (because it did not properly search for the TClass and assumed the search could not fail). | 20 January 2017, 21:23:45 UTC |
3ae9ea3 | Philippe Canal | 10 January 2017, 20:54:03 UTC | Update release notes | 20 January 2017, 21:15:43 UTC |
6d856bb | Pere Mato | 13 January 2017, 15:08:18 UTC | Update ROOT version files to v6.08/05. | 13 January 2017, 15:08:18 UTC |
79c621b | Pere Mato | 13 January 2017, 14:56:27 UTC | Updated release notes | 13 January 2017, 14:56:27 UTC |
07da889 | Pere Mato | 13 January 2017, 14:55:12 UTC | Update ROOT version files to v6.08/04. | 13 January 2017, 14:55:12 UTC |
da64a6e | Philippe Canal | 09 January 2017, 21:30:12 UTC | Remove dead code | 09 January 2017, 21:30:29 UTC |
97dd974 | Philippe Canal | 09 January 2017, 21:13:57 UTC | Do not hardcoded in which direction mmap grows the memory | 09 January 2017, 21:28:00 UTC |
1353720 | Philippe Canal | 09 January 2017, 18:04:50 UTC | Update release notes | 09 January 2017, 18:05:00 UTC |
5f6417b | Philippe Canal | 09 January 2017, 17:53:39 UTC | Allow reading of old TTreePerfStats schema. Fix ROOT-8520. Updating the schema version (ClassDef) of TTreePerfStats had been ommitted when it switched from a pure debugging (hence ephemeral) tool to a somewhat longer term utility. | 09 January 2017, 18:05:00 UTC |
9d84ab6 | Philippe Canal | 09 January 2017, 17:36:54 UTC | Fix ROOT-8523: exception catching in jitted code. Another manifestation of this problem is that, in the broken case, the stack trace shown by gdb (or any other tools) will appear corrupted. The problem appears when the memory allocated for 2 or more module happen to interleaves. This could happens because the memory needed for large module is acquired in chunks and the attempt to keep it close together (giving a hint to mmap) is not enough, since it has not way to guarante there is no hole nor that memory in between we later be release (and then re-used for a subsequent module's code). In more details: It turns out that the piecemeal allocation of memory for the code segment produced by the JIT can lead to a situation, highly dependent on memory usage of the current process and interleaving of JIT actions vs user allocations, where the code segment for multiple llvm JITed module can be interleaved. Where this causes a problem is that the code in libgcc that deal with stack unwinding (in particular the function Unwind_Find_FDE), assumes that the code segment for the modules/libraries are *not* overlapping/interleaved and thus is testing against just the start address of the code segment for the modules/libraries assuming it is enough to search for the right module for a given 'current instructions address'. In the particular case reported by Chris, one module, corresponding to the user scripts, ends up being composed of 906 code segments (each requires an individual memory allocation). There are also several small modules, one for each of the TFile plugin handler routines. In the failing case, the 906 code segments happens to be allocated in (at least) two separate memory areas (around 0x7ffff087e000 and around 0x7ffff729ef50) while the code segment for P020_TRFIOFile ends up in 0x7ffff729a000 (in the middle of the other two addresses). The user code has the simple structure: - cling wrapping function - gallery_t user function containing a try catch - parameterSetID user methods, inlineable - gallery::Handle<art::TriggerResults>::throwHandleWhyFailed_() which throw an exception. When Unwind_Find_FDE walks up the stack and reaches gallery_t, the instruction address is 0x7ffff72a2d92 and the searching algorithm then thinks that this correspond to a function in the P020_TRFIOFile module and then proceed to discover that there is no such function and no exception handler and thus 'ends' the search (and terminate the program). | 09 January 2017, 17:40:15 UTC |
4bfb09a | Axel Naumann | 21 December 2016, 15:13:29 UTC | Warn about missing support for TEntryList (until 6.10). | 21 December 2016, 15:13:29 UTC |
dada376 | Axel Naumann | 21 December 2016, 15:13:04 UTC | Improve doc and param names for SetEntriesRange(), clarifying it exludes the upper entry number. | 21 December 2016, 15:13:04 UTC |
b76a765 | Axel Naumann | 21 December 2016, 15:11:52 UTC | "Pre-deprecate" SetLastEntry(); update param names for the range. | 21 December 2016, 15:11:52 UTC |
7ea65c4 | Axel Naumann | 21 December 2016, 14:59:46 UTC | Fix & simplify example code in class reference. | 21 December 2016, 14:59:46 UTC |
bf9d821 | Lorenzo Moneta | 21 December 2016, 10:49:54 UTC | Fix check for weighted histograms | 21 December 2016, 10:50:42 UTC |
33ae299 | Lorenzo Moneta | 21 December 2016, 10:06:32 UTC | Improve check in Efficiency for weighted histograms. Fix for ROOT-8409. | 21 December 2016, 10:30:21 UTC |
c24d9c9 | Olivier Couet | 02 November 2016, 09:57:26 UTC | typos and formatting | 21 December 2016, 10:29:59 UTC |
210ee36 | Pere Mato | 16 December 2016, 11:27:15 UTC | Fix for ROOT-8510 - CMake: default config on MacOS fails due to fortran | 16 December 2016, 11:28:10 UTC |
9258a28 | Lorenzo Moneta | 15 December 2016, 15:46:36 UTC | Fix Formula for missing parameters, ROOT-8182, e.g. TFormula(“f”,“[0]+[2]”) has now 3 parameters and not 2. The second parameter, “p1” is a ghost one. This restores backward compatibility with old TFormula. Add also a test | 16 December 2016, 10:16:50 UTC |
09c644b | Philippe Canal | 11 December 2016, 20:29:51 UTC | Avoid warning about overload hiding. | 15 December 2016, 15:20:48 UTC |
d4f565a | Philippe Canal | 11 December 2016, 18:41:38 UTC | Avoid warning about overload hiding | 15 December 2016, 15:18:46 UTC |
f2e3199 | Pere Mato | 15 December 2016, 13:51:11 UTC | Fix for ROOT-8480 - CMake: root-config does not get updated | 15 December 2016, 13:52:57 UTC |
7a6e92a | Philippe Canal | 09 December 2016, 16:30:27 UTC | Re-add omitted backward compatible interface for Read/WriteStdString. Fixes ROOT-8502 | 09 December 2016, 16:31:17 UTC |
e161d03 | Axel Naumann | 06 December 2016, 10:50:37 UTC | Coding convention. (cherry picked from commit 0f5f489d66f29b0baa393baf699f147ecb91ac78) | 09 December 2016, 08:47:09 UTC |
c0dc651 | Axel Naumann | 06 December 2016, 10:01:23 UTC | Set fImpl even if proxy exists. Split giant CreateProxy(). See https://root.cern.ch/phpBB3/viewtopic.php?f=3&t=22790. (cherry picked from commit 3c618d2e063b55c34e358cdf7e88469ad2276198) | 09 December 2016, 08:47:03 UTC |
db71f77 | Axel Naumann | 06 December 2016, 09:11:25 UTC | Missing fImpl is not a reason to abort(). (cherry picked from commit 252776a3d80b59a687d057dc587a1c433f3d3cf0) | 09 December 2016, 08:46:57 UTC |
597d4cb | Axel Naumann | 08 December 2016, 19:02:30 UTC | Only get the init value if clang has it (ROOT-8499). (cherry picked from commit e8df1bd4a32beb2c38774b2227a8db5f842a8c86) | 09 December 2016, 08:44:00 UTC |
e124d50 | Pere Mato | 07 December 2016, 22:31:36 UTC | Fixed 'make distsrc' when run in isonation | 07 December 2016, 22:36:50 UTC |
32908f4 | Lorenzo Moneta | 07 December 2016, 16:39:12 UTC | Fix missing parenthesis when handling polynomial. Fix ROOT-8496 Add corresponding test | 07 December 2016, 16:39:12 UTC |
1aa75dd | Lorenzo Moneta | 07 December 2016, 14:58:16 UTC | Fix option I (intergral of function in bin) for TH1::Add(TF1) . See https://root.cern.ch/phpBB3/viewtopic.php?f=3&t=22816 | 07 December 2016, 15:06:48 UTC |
37b34ff | Pere Mato | 05 December 2016, 21:11:31 UTC | Upated release notes | 05 December 2016, 21:11:31 UTC |
89cfad1 | Philippe Canal | 05 December 2016, 15:58:34 UTC | Fix ROOT-8490, avoid nullptr point to std::string. Also improve handling of the case of ~invaliduser by leaving the filename as is and allow TFile::Open to print the original filename in the error message. | 05 December 2016, 16:00:56 UTC |
b5b59e1 | Philippe Canal | 02 December 2016, 23:11:27 UTC | Update ROOT version files to v6.08/03. | 02 December 2016, 23:11:27 UTC |
dd4fcde | Philippe Canal | 02 December 2016, 23:07:29 UTC | Add part of v6.08/02 release notes | 02 December 2016, 23:07:29 UTC |
7f705b8 | Philippe Canal | 02 December 2016, 22:48:16 UTC | Update ROOT version files to v6.08/02. | 02 December 2016, 22:48:16 UTC |
9f7b656 | Philippe Canal | 02 December 2016, 19:48:16 UTC | Still need to copy ClingRAII.h into include dir | 02 December 2016, 19:48:16 UTC |
faaa76d | Philippe Canal | 02 December 2016, 18:01:14 UTC | Revert "Prompt error while building StreamerInfo if element is vector<T> and has no dictionary" This reverts commit bcec6aee7e2077862d606b40290971c345927dd3. | 02 December 2016, 19:10:57 UTC |
8aa22e9 | Philippe Canal | 02 December 2016, 17:16:01 UTC | Remove the (incorrect) forced alignment of the content of vector. This was the 'right' thing to do for Emulated class (due to a deficiency in TStreamerInfo::ComputeSize) but wrong from loaded classes (where TClass::Size() is authoritative for use in vector). Which this change we should be back to the emulated vector and a compiled vector being bitwise identical. Note the forced alginment was introduced in 9968d0f0301776294f361752ff7b914e7c181f58 to more properly support emulated class (which is more comprehensively done by fixing TStreamerInfo::ComputeSize) | 02 December 2016, 19:10:57 UTC |
c56dcee | Philippe Canal | 02 December 2016, 17:11:04 UTC | Use constexpr and mark variable name as const. Avoid code duplication | 02 December 2016, 19:10:56 UTC |
6281247 | Philippe Canal | 02 December 2016, 17:10:36 UTC | Align the 'sizeof' emulated classes. This avoid user of this size (like the emulation of std::vector) to have to align it (and having to filter this alignment on just the emulated case) | 02 December 2016, 19:10:56 UTC |
d0e3c08 | Philippe Canal | 02 December 2016, 17:08:51 UTC | Add more code comment | 02 December 2016, 19:10:56 UTC |
b0220d6 | Philippe Canal | 02 December 2016, 17:08:38 UTC | Use constexpr and mark variable name as const | 02 December 2016, 19:10:56 UTC |
77e1f4d | Philippe Canal | 01 December 2016, 16:38:46 UTC | ClingRAII.h should not be in the dictionary | 01 December 2016, 16:39:17 UTC |
4c6d03c | Axel Naumann | 30 November 2016, 15:28:56 UTC | Revert "Set -fvisibility-inlines-hidden throughout ROOT." This breaks roottest/python/cpp/PointerPassing.C This reverts commit 676d2b2ab3f98e7f1bef9d69292ca014dd430945. | 30 November 2016, 15:28:56 UTC |
5ece58a | Axel Naumann | 30 November 2016, 15:28:13 UTC | Revert "As for MacOS, hide inlines." Breaks roottest/python/cpp/PointerPassing.C. This reverts commit 39090cb364611d41eae5ebafb930a50615d30cb2. | 30 November 2016, 15:28:13 UTC |
e1d6f3b | Enrico Guiraud | 30 November 2016, 11:15:31 UTC | Change "TTreeReader::Reset" to "Restart" in error message (cherry picked from commit 8af4698e513d4ebebdb3b272a93e93fa7fa1e3cb) | 30 November 2016, 12:01:52 UTC |
cadd188 | Axel Naumann | 29 November 2016, 12:43:33 UTC | The dtor shall be public for polymorphic delete. Thanks, Enrico! (cherry picked from commit f48823c0ce8a121672c377e3a99974374ff90f97) | 29 November 2016, 13:40:38 UTC |
39090cb | Axel Naumann | 29 November 2016, 13:37:46 UTC | As for MacOS, hide inlines. (cherry picked from commit 5beed7bec9514eee573febc23cd202963e59664f) | 29 November 2016, 13:38:36 UTC |
325eb19 | Gerardo Ganis | 29 November 2016, 09:41:29 UTC | Update release notes | 29 November 2016, 09:41:29 UTC |
4a90a84 | Gerardo Ganis | 22 November 2016, 10:24:27 UTC | proof/proof: fix for ROOT-8466 Remove residual symlink creation (after optimization introduced in 0a976f4a56cf55bd8102c5d89088531be874ef1a). For PROOF-Lite, make sure that all the relevant files (including .pcm) are present in the cache. | 29 November 2016, 09:36:42 UTC |
66b6682 | Gerardo Ganis | 22 November 2016, 10:11:40 UTC | proof/proof: silence warning from TUri | 29 November 2016, 09:36:42 UTC |
f14c9e1 | Gerardo Ganis | 22 November 2016, 10:10:30 UTC | proof/proofplayer: consolidate TProofPlayer::AssertSelector | 29 November 2016, 09:36:42 UTC |
6b96095 | Danilo Piparo | 29 November 2016, 09:24:50 UTC | Release notes | 29 November 2016, 09:24:50 UTC |
dc17d27 | Danilo Piparo | 29 November 2016, 09:23:16 UTC | Release notes | 29 November 2016, 09:23:20 UTC |
2cfbc66 | Danilo Piparo | 29 November 2016, 09:21:41 UTC | Release notes | 29 November 2016, 09:21:51 UTC |
d8d5167 | Axel Naumann | 29 November 2016, 08:51:53 UTC | Add -patches section; mention -fvisibility-inlines-hidden. | 29 November 2016, 08:52:02 UTC |
676d2b2 | Axel Naumann | 29 November 2016, 08:40:23 UTC | Set -fvisibility-inlines-hidden throughout ROOT. Avoids "different visibility settings" warning. (cherry picked from commit df0a8c4b6415a23752b035b5785faa8ddb6c5927) | 29 November 2016, 08:52:02 UTC |
c6741f7 | Danilo Piparo | 29 November 2016, 08:48:15 UTC | Prompt error while building StreamerInfo if element is vector<T> and has no dictionary This fixes ROOT-8478. | 29 November 2016, 08:50:32 UTC |
f0de158 | Axel Naumann | 28 November 2016, 16:50:25 UTC | Hide the most absurd CPP macros from mmalloc headers (ROOT-8475). Not needed in the master because that uses res/ and hides mmalloc. | 28 November 2016, 16:50:25 UTC |
b359b7f | Philippe Canal | 28 November 2016, 15:26:13 UTC | ClingRAII.h should not be in the dictionary | 28 November 2016, 15:26:13 UTC |
18c7013 | Philippe Canal | 25 November 2016, 19:14:08 UTC | Factor ParsingStateRAII and use in TClingCallbacks. Aslo add DelayedInfoRAII | 25 November 2016, 19:15:13 UTC |
1640f03 | Philippe Canal | 22 November 2016, 21:22:45 UTC | Follow spirit of naming convention (_t is for typedef). _t for struct is a renmant of the fact that C sorta requires the use of typedef for struct naming | 25 November 2016, 19:15:13 UTC |
fa5daff | Philippe Canal | 25 November 2016, 16:19:17 UTC | Store an empty StreamerInfo if the file has no data or data that does not need a StreamerInfo. This avoid spurrious warning messages | 25 November 2016, 17:50:40 UTC |