Should `concurrent` status be passed on along with group_by?

[gaow]

Currently with the new behavior of group_by what I used to specify for a workflow,

[1]
input: group_by = 1, concurrent = True
...
[2]
input: group_by = 1, concurrent = True

now becomes

[1]
input: group_by =1 , concurrent=True
...
[2]
input: concurrent=True

I wonder where information for concurrent is saved, and if it makes sense to make it group property that can also be passed on unless otherwise specified?

[BoPeng]

The only case when concurrent should be False is when the next substep depends upon the result of previous ones, and frankly I have not seen a single use case for this in the past few years. With the better control of number of processes now, I suppose we can make concurrent=True the default.

[gaow]

@BoPeng just to double check this: now concurrent=True is default for non-task. It is also True for task, and has been like that for some time already, right?

[BoPeng]

Yes. I do not think there is such an option for task (and I will remove it if there is as I am going through the manual).