Peter Willendrup

Related? It looks like the Python interpreter may have been overwhelmed by lots of text output from the many MPI cores...

(Here from 32 processes on DTU windows box)

@farhi discussion most welcome.

@farhi I have now implemented a prototype which simply copies back x, y, z, .... p inside a function. This should leave USERVARS and other logic parameters alone. Let's see...

After a pretty awful typo on the McStas side, the test last night gave lots of errors: http://new-nightly.mcstas.org/latest/2022-03-25_output.html - a rerun after correction puts most things right, except: * ILL_D2B...

The common denominator in the diff's is very likely related to Monitor_nD restore_neutron=1 which seems affected to the new way of restoring... (Shadow on the "non-scattered" part of the beam??)...

... more specifically seems related to cylinder geometry "lid and bottom"

Looks like this is likely closer connected to the radial collimator / non-transported particles

This has all fallen into piece with the latest GROUP/restore related edits

It seems the issue is specific to that function / instrument, this one works: [dataset.zip](https://github.com/McStasMcXtrace/McCode/files/8976567/dataset.zip) [Test_Magnetic_Constant2.instr.txt](https://github.com/McStasMcXtrace/McCode/files/8976568/Test_Magnetic_Constant2.instr.txt)