[R-Forge #5674] setDT + microbenchmark interaction loses over-allocation somehow #473
Description
Submitted by: Arun ; Assigned to: Nobody; R-Forge link
As reported here on SO.
The comment clearly explains how to reproduce the issue. Naive check shows that microbenchmark calls the function to evaluate to .Call. And this doesn't seem to assign the over-allocated data.table back to its parent environment correctly for some reason and therefore leads to warning later when using := that a copy (which is actually a shallow copy) has been taken.