> On 18 Feb 2021, at 04:09, Thomas Munro <thomas.mu...@gmail.com> wrote:
> In another thread[1], I proposed $SUBJECT, but then we found a better > solution to that thread's specific problem. The general idea is still > good though: it's possible to (1) replace several existing copies of > our qsort algorithm with one, and (2) make new specialised versions a > bit more easily than the existing Perl generator allows. So, I'm back > with a rebased stack of patches. I'll leave specific cases for new > worthwhile specialisations for separate proposals; I've heard about > several. Just to play around with this while reviewing I made a qsort_strcmp, like in the attached, and tested it using a ~9M word [0] randomly shuffled wordlist. While being too small input to make any meaningful difference in runtime (it shaved a hair off but it might well be within the error margin) there was no regression either. More importantly, it was really simple and quick to make a tailored qsort which is the intention with the patch. While still being a bit of magic, moving from the Perl generator makes this slightly less magic IMO so +1 on this approach. A tiny nitpick on the patch itself: + * - ST_COMPARE(a, b) - a simple comparison expression + * - ST_COMPARE(a, b, arg) - variant that takes an extra argument Indentation. All tests pass and the documentation in the the sort_template.h is enough to go on, but I would prefer to see a comment in port/qsort.c referring back to sort_template.h for documentation. -- Daniel Gustafsson https://vmware.com/ [0] https://github.com/dwyl/english-words/ shuffled 20 times over
qsort_tmpl_strcmp-patch
Description: Binary data
