[RF] RooFit workspace work differently with pyROOT in 6.32-rc1 version and 6.30/02 about root HOT 2 OPEN

Hi @yuhao-wang-nju, thanks for your post!

1. This is by design (see #15153)
2. Where does the doc say that? I think cppyy pythonizes the cppyy.gbl.std.string such that it behaves like a Python str, therefore it can be used in many contexts where one would use str. But this doesn't always help. The type is different after all, and there is no automatic conversion for the reasons outlined in #15153. You will have to do this yourself with str() around the std::string.
3. The evaluation was completely replaced in 6.32. Probably you have one of the corner cases where the new backend is slower and I need to debug this. Is it also possible to reproduce this slow fit with the workspace you shared? Maybe you can share the full script?

Thanks a lot!

Related PR: #14237

from root.

Hi @guitargeek , thanks a lot for the reply and sorry for the late message.

For the first and second point, thanks a lot for the explanation. I may misunderstood the way cppyy works previously, now it seems quite reasonable to me!

For the third one, I prepared a workspace and also attached the code that we use (The dataset named obsData contained in the file is an asimov one). We are using a framework named WSMaker and I just picked up the fitting code. It is a bit long and hope it won't bring too much trouble to you. Here is the command we use for fit:
root -l -b -q '$WORKDIR/FitCrossCheckForLimits.C+(FitToAsimov,1.0,0.0,true,"ws.root","output/","combined","ModelConfig","obsData",false,false)'

The file is a bit large so here is the gdrive link. Thanks a lot in advance for your time!

from root.