New nu stack for NOvA art2 --> art3 transition
Can we get a nu stack with
and if we can get this too, it would be great:
#3 Updated by Lynn Garren 3 months ago
Be aware that you will be getting a nearly complete set of new products. The table lists highlights.
|NOvA current||LArSoft current||NOvA with art v3|
|python||v2_7_15a||v2_7_15a and v3_7_2||v2_7_15a and v3_7_2|
#6 Updated by Brian Rebel 3 months ago
Thanks for providing the list of versions. Is it possible to get nug4, geant4reweight, nugget and nutools to have the same versions in the new stack as the current NOvA stack?
Also, can you tell us what changed in nusimdata between v1_16 and v1_21? Were those changes just related to art2-->art3, or is there something else?
#7 Updated by Lynn Garren 3 months ago
nusimdata has changes required by updating art/canvas. Also, because of changes in underlying products and added support for e19, you are unlikely to get the same version of anything. The questions in the table are simply because we don't yet know exactly how that will play out. The table will be updated as I go along.
#8 Updated by Brian Rebel 3 months ago
I see. If we can keep the versions of GENIE and GEANT4 the same (modulo an identifier for the compiler) that would be most helpful in our vetting of the new product stack. It would cut down on the number of places we have to look when the inevitable differences arise.
#10 Updated by Brian Rebel 3 months ago
OK, thanks. Yes, I know the table file changes. It wasn't clear that you meant we were getting the same versions of the underlying product from the previous comments. If I understand correctly now, you mean the ups product versioning is changed, ie the letter appended at the back of the version, but not the underlying release of the package. So, for example GENIE will remain at version v3_00_06_p01 from the GENIE repository, even though the ups product might have a version of v3_00_06_p01a. Am I understanding you correctly?
#11 Updated by Alexander Himmel 3 months ago
I will second Brian’s point that we want to avoid changing any “physics” here if at all possible since that opens up substantial new validation requirements.
However, from the discussion it sounds like that’s not expected, and we’re just changing our internal markers to make them unique products, right?
#17 Updated by Lynn Garren 2 months ago
- % Done changed from 0 to 80
nulite v3_08_00 is now available. The table above has been edited to reflect the versions of products included.
We are still waiting for cstxsd, which will be part of a matching nu distribution.
art v3_05_01 only supports e19 and c7 builds. The python2 builds use the py2 qualifier.
With clhep 2.4, the default random number generator changed. If you always specify the generator, this is not a problem.