Incompatibility with earlier ROOT versions

[rumashan]

Incompatibility with earlier ROOT versions

[VinzenzBildstein]

I added the changes (plus a few minor improvements and bug fixes). This should solve this issue, but the automatic closing done by GitHub might have been too early, so I re-opened it for now.
@rumashan can you please update your GRSISort and GRSIData to the newest versions and check that those compile on your system as well?

[VinzenzBildstein]

Thank you for reporting these issues. The problem with using functions not available in older root versions is an old one, I think this will keep coming up until I can find an easy way to do test compilations with different ROOT versions. I am working on making GRSISort build with cmake, which should enable us to do this as a part of the GitHub actions whenever someone creates a pull request.
I'm a bit surprised by the problems with the wigner_6j function, because I did encounter that problem recently and added another check for the HAS_MATHMORE variable around it. But you are saying that your root version does have mathmore installed (meaning root-config --has-mathmore returns yes)?

[rumashan]

Thanks for your response, Vinzenz. I double checked like you suggested. root-config --has-mathmore does return yes.

[VinzenzBildstein]

I just checked this, and noticed the actual error for the wigner_6j function is because it tries to use TGRSIFunctions::ROOT::Math. I think this might be due to the NamespaceImp macro called at the beginning of the source file, but I'm not sure as I didn't find any documentation of that having been changed (and doing a vimdiff between the Rtypes.h files in 6.08.06 and 6.30.02 doesn't show a big difference).
But I moved the definition of that function to the header, and now that part at least compiles under 6.08.06.
I will check the other errors this is throwing and push all the changes once that version compiles as well.

[rumashan]

Thanks Vinzenz. I'm being curious and this is possibly irrelevant - In the same source file TGRSIFunctions.cxx, the LegendrePolynomial() function also uses ROOT::Math but on compilation, it doesn't throw an error. If the issue is with the NamespaceImp macro, shouldn't this also throw errors?

[VinzenzBildstein]

You are right. Careful comparison of the two statements shows that the one in LegendrePolynomial starts with ::ROOT::TMath, whereas the statement in the wigner_6j function was just ROOT::TMath. The leading colons ensure that this namespace starts in the global namespace, not the current one. With that change (leaving the definition in the source file and adding the leading colons) and a few others, GRSISort compiles for me using ROOT 6.08.06. Next step is to get GRSIData to compile as well (and double-check that everything still compiles with a newer version).

[rumashan]

Ah gotcha, that makes sense. Thanks for looking into GRSIData as well.

[rumashan]

I checked your recent changes and both GRSISort and GRSIData compiled successfully. Thank you Vinzenz!

PS. The GRSISort compilation had a couple of minor warnings. In case you're interested in seeing what they are, I've attached my console output.
grsisort-compile.txt

[rumashan]

Before this gets closed, I want to bring up something possibly related. I'm having some issues with running the standard ExampleEventHelper using GRSIFrame. I'm testing it on analysis20323_000.root which is an in-beam run from S1723 (Aug-2021) beamtime. I unpacked this run using the latest version of GRSISort
The Helper compiles successfully but crashes promptly after that. The output .root file does not contain any histograms. I don't see anything overtly wrong with the Helper or my Analysis Tree and I'm at a loss to what could be causing this. Apologies if it's a separate issue and should've been opened in a new thread.

Here is the terminal output after running grsiframe and also the output of ErrorReport.sh.
grsiframe-log.txt
error-report.txt