Make sure the saved TGraph is a proper C++ variable

[couet]

Make sure the saved TGraph is a proper C++ variable. As suggested here: https://its.cern.ch/jira/browse/ROOT-8277 it works also for the classes inheriting from TGraph

[github-actions[bot]]

Test Results

    11 files      11 suites   2d 14h 17m 21s :stopwatch:  2 644 tests  2 644 :white_check_mark: 0 :zzz: 0 :x: 27 444 runs  27 444 :white_check_mark: 0 :zzz: 0 :x:

Results for commit 7c9b975e.

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

[couet]

Indeed it is just using gInterpreter->MapCppName if a test should be done it would be to test MapCppName. I thought a bit about it and I think this issue might be also seen with some other objects like histograms. I plan to check that.

[dpiparo]

Is mangling supposed to address those characters too? Maybe there is something simple which could be done for names of objects in ROOT, for example keeping only letters and numbers and postponing the hash of the name? In the end we need this for autogenerated code, but maybe I am wrong.

[couet]

Indeed this approach using gInterpreter->MapCppName is already used in some other SavePrimitives methods (like TH1). Seems to me it makes sense to continue with the same technique in the class where it is missing. I noticed some characters were missing @devajithvs is looking at it.

[dpiparo]

Hi @couet , are we still missing any ancillary developments to proceed with this PR?

[couet]

I am on it. It is more complex than expected. TH2Poly is more complex