[DF] Allow calling RunGraphs and VariationsFor on distrdf results

[eguiraud]

Companion roottest PR at https://github.com/root-project/roottest/pull/980 .

[github-actions[bot]]

Test Results

         9 files           9 suites   1d 20h 28m 56s :stopwatch:   2 472 tests   2 465 :heavy_check_mark: 0 :zzz:   7 :x: 21 147 runs  21 129 :heavy_check_mark: 0 :zzz: 18 :x:

For more details on these failures, see this check.

Results for commit 192143fc.

:recycle: This comment has been updated with latest results.

[eguiraud]

With this patch, this code:

import ROOT
from dask.distributed import Client, LocalCluster

if __name__ == "__main__":
    client = Client(LocalCluster(n_workers=1, processes=False, threads_per_worker=1))
    df = ROOT.RDF.Experimental.Distributed.Dask.RDataFrame(10, daskclient=client)
    df = ROOT.RDataFrame(10)
    c = df.Count()
    cs = ROOT.RDF.Experimental.VariationsFor(c)
    print(cs["nominal"])

breaks with this error:

Traceback (most recent call last):
  File "/tmp/distrdf.py", line 9, in <module>
    cs = ROOT.RDF.Experimental.VariationsFor(c)
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/blue/ROOT/master-with-venv/cmake-build/install/lib/ROOT/_facade.py", line 96, in variationsfor
    return rdf_variationsfor(resptr)
           ^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: Template method resolution failed:
  ROOT::RDF::RResultPtr<ROOT::RDF::RInterface<ROOT::Detail::RDF::RLoopManager,void> > ROOT::RDF::Experimental::VariationsFor(ROOT::RDF::RResultPtr<ROOT::RDF::RInterface<ROOT::Detail::RDF::RLoopManager,void> > resPtr) =>
    TypeError: could not convert argument 1
  Failed to instantiate "VariationsFor(ROOT::RDF::RResultPtr<ULong64_t>*)"

Probably some issue with monkey-patching a template function, but I don't understand what goes wrong exactly or how to fix it. Needs debugging.

[vepadulano]

I opened a PR that practically supersedes this (https://github.com/root-project/root/pull/16681). It introduces the fixes necessary for the issues highlighted here, plus also unifies the local and distributed RDataFrame constructors. We can move further conversation over to the new PR.

[guitargeek]

Superseded by the following PR, as explained in https://github.com/root-project/root/pull/13248#issuecomment-2412528571:

• https://github.com/root-project/root/pull/16681