Skip to content

Fix bugs in slices where all events fail GEC and in sub-slices

Daniel Charles Craik requested to merge dcraik_fix_gec_bugs into master

This MR fixes two bugs:

  • The GEC event_list was starting from 0 even when processing the second part of a split slice. It now adds on the value of the first element from the pre-initialised list. Note that this still assumes the event numbers are consecutive but this is always true for events entering the GEC and the more general fix would have been incompatible with the allen_tdr branch.
  • Allen was crashing if all events in a slice failed the GEC. This was because the block dimension of most algorithms is set by the number of events that pass the GEC and a dimension of zero is not allowed. We now check for zero before running the kernel.

Merge request reports