ROOT.jl: ROOT bindings for julia

Here is guidebook for people familiar with ROOT starting from scratch with julia, demonstrating some of the capabilities: https://github.com/jpata/JuliaForPhysicists/blob/master/guide.ipynb

Wraps ROOT (http://root.cern.ch) using Clang.jl in a semi-automatic way. Currently supports a subset of the most commonly used ROOT classes and methods, among them:

• TTree, TChain, TBranch
• TFile, TDirectory, TDirectoryFile
• TH1D, TH2D
• TCollection, TList, TKey

This wrapper is a very thin C layer on top of the ROOT C++ library and no runtime interpreter (e.g. CINT or Cling) is called for the core functions. Therefore, this is theoretically the fastest possible wrapper one can use short of using C++ directly.

One can access the full ROOT framework though the process_line(cmd::ASCIIString) wrapper.

See also https://github.com/jpata/ROOTDataFrames.jl for an interface between TTrees and DataFrames. See also https://github.com/jpata/ROOTHistograms.jl for an interface between https://github.com/JuliaStats/StatsBase.jl histograms and julia histograms.

Installation

You need ROOT v6 and julia v0.4. After configuring ROOT using thisroot.sh, execute the following in a julia prompt

> Pkg.clone("https://github.com/jpata/ROOT.jl.git")
> Pkg.build("ROOT")
ROOT.jl compiled!
Add the following to your ~/.bashrc or ~/.bash_profile:
alias rjulia="/Users/joosep/.julia/v0.4/ROOT/julia"

Now run the ROOT-enabled julia interpreter using /Users/joosep/.julia/v0.4/ROOT/julia

> using ROOT
> Pkg.test("ROOT")
> h = TH1D("hist", "My Hist", Int32(10), 0.0, 100.0)

This will create a new executable julia in the ROOT.jl package directory. This must be used as the julia executable if you want to simultaneously use ROOT.

Usage

Short examples are given under the example directory. In general, the use pattern is very similar to PyROOT. One should keep in mind that every TObject in julia is a pointer to an object allocated on the heap (using new in C++).

example TTrees

These are the main workhorse of ROOT, which support fast row-based and disk-backed data storage for simple data types.

#file: example/ttree.jl
#run as: rjulia example/ttree.jl
using ROOT

tf = TFile("test.root", "RECREATE")
ttree = TTree("my_tree", "My Tree")

#branch variable should be array with length 1
x = Float64[0.0]
px = convert(Ptr{Void}, x)
br = Branch(ttree, "x", px, "x/D")

for i=1:10000
    x[1] = randn()
    Fill(ttree)
end

Write(tf)
Close(tf)

Known issues

1. ROOT.jl supports only ROOT 6
2. When calling using ROOT from the standard julia REPL that comes with julia (without having recompiled as in this module), Cling crashes/segfaults: always use rjulia with ROOT.jl.
3. Only a limited set of classes are supported
4. There is no mechanism to destroy ROOT-managed objects