different pipelines in catalyst should be allowed to have the same grid name (or not?)
A fairly recent change (Nov. 2021, I believe) added a restriction on naming grids to require that no two grids have the same names. The relevant change is an addition to coprocessing.py in the method CreateProducer(self, datadescription, inputname):
if inputname in self.__ProducersMap:
raise RuntimeError("CreateProducer is being called multiple times for input '%s'" % inputname)
In the check-in comment utkarsh mentions that this is to avoid subtle errors down the road.
However, this broke our Sierra/Catalyst installation, because we use a default name of "input" and can sometimes have multiple "result output" sections which result in different catalyst pipelines and different datadescriptions.
Rather than restricting the names, we could have __ProducersMap be a two-level map with the datadescription as the first key and the name as the second key. Or __ProducersMap could reside in datadescription. Or we could change our Sierra code to give a different default name to each section and make the script on the other side robust to this change.
__ProducersMap not keying on datadescription caused another problem with multiple grids for multiple pipelines--it may not make sense to use the input grid name as a unique key. Or it might.
We need to talk about this before we do anything in particular (or leave it as is).