Repository navigation
Nested loops fail #340
Description
The following code fails:
import flowrep as fr
def arange(min_value, max_value):
result = range(min_value, max_value)
return result
def sub(a, b):
result = a - b
return result
def mean(values):
result = sum(values) / len(values)
return result
@fr.workflow
def get_mean_value(min_value=0, max_value=3, x=2):
r = arange(min_value=min_value, max_value=max_value)
outer_result = []
for v in r:
inner_result = []
for w in r:
b = sub(w, x)
inner_result.append(b)
inner_mean_value = mean(inner_result)
a = sub(v, inner_mean_value)
outer_result.append(a)
mean_value = mean(outer_result)
return mean_valueReacted by Copilot
Activity
liamhuber commentedon Oct 22, 2026
I'm copying your comment here, because #341 itself is A-OK from a technical perspective and I don't want to clutter that.
I did a quick check of my gut hypothesis, and this really is fundamentally only a label tracking issue; if we reassign a fresh symbol to the list before reusing it, the code parses and runs completely fine:
So if it would run with a trivial relabel, I definitely consider it a bug that it won't run without one.
To directly address your concern, in this case writing arange as a list of lists would be memory wasteful, since each sub-list would be identical. We can make the inner list depend on the outer list by replacing
b = sub(w, x)withb = sub(w, v), at which point it becomes meaningful and interesting to have made a nested iteration over the same source data twice.At the end of the day, I'm coming down solidly on the side that this is a real parsing bug and we might as well patch it.