why comparing with quicksort?

[andrewmed]

It seems that the benchmark is comparing with GO's implementation of quicksort (which is standard in go)

timsort/bench_test.go

Line 143 in 13f2155

sort.Sort(&v)

Given that timsort is a stable sort, would not it be right to compare with sort.Stable() instead?

I got much worse results with that:

RevSorted100: 1500          (7806)        1582          (7759)        1469          (7773)        
Xor100:       4838          (5802)        4828          (5845)        4819          (5834)        
Random100:    5244          (7891)        5250          (7887)        5242          (7911)        

Sorted1K:     5597          (4837)        5584          (4838)        5633          (4846)        
RevSorted1K:  6588          (87094)       6523          (85785)       6562          (88508)       
Xor1K:        45714         (114406)      45331         (113812)      46041         (114143)      
Random1K:     114125        (202941)      132139        (203927)      114355        (202147)      

Sorted1M:     4129627       (8783352)     4129406       (8771254)     4148070       (8678536)     
RevSorted1M:  6239082       (114131971)   6189573       (115857436)   6199218       (114664717)   
Xor1M:        99852677      (224848697)   98523482      (228193397)   100662027     (222984886)   
Random1M:     362650134     (665379425)   348801121     (669130321)   340476927     (670888160)```

[psilva261]

That's true! Do you want to make a pull request? Otherwise I'll make one when I find the time, probably also adding that in the README.

(One time in the future I might add a GNU plot to this)