In the cases it's supposed to address, it does make a striking difference. I interrupted a bulk build that was about 1000 packages along.
The I restarted it twice, once without -d, once with -d. Without -d, there is a flat queue at start, and the jump when vars discovers groff is striking. With -d, the queue grows at a fast, mostly linear rate while LISTING progresses. (attached graphs)
<<attachment: 1.png>>
<<attachment: 2.png>>
