random() cached across rows when inside list comprehension #8108
Closed
Description
What happens?
The value for random() repeats across rows when used inside a list comprehension.
To Reproduce
D select [random() for _ in range(1)] from range(2);
┌──────────────────────────────────────────┐
│ main.list_apply(range(1), _ -> random()) │
│ double[] │
├──────────────────────────────────────────┤
│ [0.765232979785651] │
│ [0.765232979785651] │
└──────────────────────────────────────────┘
The two rows are always the same. Contrast that with:
D select random() from range(2);
┌────────────────────┐
│ random() │
│ double │
├────────────────────┤
│ 0.5257333947811276 │
│ 0.5609628073871136 │
└────────────────────┘
D select [random()] from range(2);
┌───────────────────────────┐
│ main.list_value(random()) │
│ double[] │
├───────────────────────────┤
│ [0.3676034810487181] │
│ [0.6335314936004579] │
└───────────────────────────┘
Including a value that changes in the list comprehension fixes the issue:
select [random() + range * 0 for _ in range(1)] from range(2);
┌────────────────────────────────────────────────────────────┐
│ main.list_apply(range(1), _ -> (random() + ("range" * 0))) │
│ double[] │
├────────────────────────────────────────────────────────────┤
│ [0.25919290888123214] │
│ [0.4220627495087683] │
└────────────────────────────────────────────────────────────┘
OS:
MacOS
DuckDB Version:
v0.8.2-dev161 563662b
DuckDB Client:
CLI
Full Name:
Paul Rosenzweig
Affiliation:
None
Have you tried this on the latest master branch?
- I agree
Have you tried the steps to reproduce? Do they include all relevant data and configuration? Does the issue you report still appear there?
- I agree