Haskell-perf/sequences shows interesting results on addition of DList, Acc & snoc into comparison

Now old DList suggestions/claims regained steam with real proof numbers backing them up:

The target audience here would be interested to have on hand tangible comparative DList benchmark numbers.

That results of benchmarks are proper - were rechecked with spl, (author of DList) & haskell-perf author.


P.S.

Strongly suggesting not forcing strictness on DList.

Reddit:

1 Like

Graphs would be helpful! The tables are hard to interpret at a glance.

1 Like

This is not a place to ask for it, it is a nice thing to open a report to ask in the project & implement it in the project.