I have written the code now and run a benchmark. In the benchmark function slices.Sort is 9% faster than slices.SortFunc, which is ca. 9% faster than slices.SortStableFunc for my benchmark. I don't need stable sort, but assumed that merge sorting would benefit more from presorted runs.
On Sunday, April 23, 2023 at 8:57:59 AM UTC+2 Axel Wagner wrote: > It has not been added because it is pointless. Stable sorting is generally > slower than regular sorting, as it puts extra constraints on the output and > keeping to those makes things slower. And for a total order, *every* > sorting algorithm produces a stable sort, as stability only matters if you > have distinguishable but not ordered elements, which can't happen for total > orders. Discounting NaN and ±0 issues, < is a total order and it doesn't > seem worth it to add a new API for the sole purpose of promising to sort > NaNs stably. > > On Sun, Apr 23, 2023 at 8:05 AM Uli Kunitz <uli.k...@gmail.com> wrote: > >> I have a use case where I have to sort int32 slices repeatedly that are >> already partially sorted. (Imagine a tree of smaller non-overlapping >> segments of the same slice, where sorting starts at the bottom and moves to >> the top and the results of the individual sorts are required for the >> algorithm.) Speed is critical for the use case. The StableSort method is an >> excellent first approach for it, because it uses the SymMerge algorithm and >> for small blocks InsertionSort benefiting from pre-sorted runs. Small >> performance improvements benefit the use case, so I looked at >> golang.org/x/exp/slices which contains a SortStableFunc but not a pure >> SortStable method, which I would like to use because I need only the order >> implied by the default less-than (<) operator. For non-stable sorting there >> is pure Sort function, so I wonder why SortStable has been left out. Has it >> not been added because of the NAN issue? >> >> For now I can live with SortStableFunc since I plan a custom sort using >> the knowledge of the positions of the sorted sub-segments, but I'm puzzled >> about the lack of a pure SortStable. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "golang-nuts" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to golang-nuts...@googlegroups.com. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/golang-nuts/89ad151a-a604-45d5-a6a4-620c0024060an%40googlegroups.com >> >> <https://groups.google.com/d/msgid/golang-nuts/89ad151a-a604-45d5-a6a4-620c0024060an%40googlegroups.com?utm_medium=email&utm_source=footer> >> . >> > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/6f86b936-3149-4f31-800e-0b69cd96a6c5n%40googlegroups.com.
