First release of the v6.32 series.
🗒️ Release notes
💾 Install instructions
Highlights:
- RDataFrame - zero-code-change experience when moving from processing a TTree to processing an RNTuple, as well as a greatly improved Distributed RDataFrame: ROOT is ready to run at your favourite Analysis Facility – it even allows you to profit from an interactive experience backed by a distributed system using your current batch system (e.g. HTCondor, like the lxplus+lxbatch combination at CERN): try it now!
- RooFit - The new vectorizing CPU evaluation backend is the default for likelihood minimization, now up to 10x faster on a single CPU core!
- PyROOT - the interop engine of PyROOT, cppyy, was upgraded to its latest version, blurring the boundaries between Python and C++ in ROOT better than ever, e.g. the conversion of NumPy arrays to vectors, implicit conversion from nested Python tuples to nested initializer lists, and improved overload resolution.
- RNTuple – The RNTuple on-disk format was updated to release candidate 2, in preparation of the binary format first production freeze. The RNTuple API come with a major refactoring, improving consistency across different parts and improving overall robustness. Moreover:
- Merging of RNTuple data with hadd is now supported.
- A new RNTupleParallel writer class creates RNTuple data in highly concurrent settings.
- A new RNTupleInspector utility class provides information about the on-disk metadata of an RNTuple.
- Cling - Speaking of interpreters, the new PyROOT is glorified by a new LLVM version, LLVM 16, that comes with numerous advantages, among which a better support for C++ 20 as well as better and faster generated code.
- Graphics - The ROOT release 6.32 brings a lot of impressive enhancements to the Web Graphics package, surpassing the features and capabilities of version 6.30. This update provides users with a secure and more robust Web Graphics. Try it with the command
root –web! - REve - Lightweight rendering of any shape on the scale of hundred thousand of instances. The box, hexagon, and cone shape are showcased in the eve7 examples. The rendering of transparent objects is greatly improved.
All that comes with a greatly improved stability: more than 250 items in the ROOT trackers have been addressed for this release. Excellent news for experiments planning to include this release in their production software stacks!