Closed rainhead closed 13 years ago
Patrick Galbraith is working on this. Maybe you could coordinate with him to help with the patch?
Hmm, benchmarks suggest that 0.47 is a performance regression. :-( So I am gonna leave this alone. You can checkout my partially working branch '0.47' if you want.
0.32:
set: libm:ascii 0.610000 0.970000 1.580000 ( 3.654520)
set: libm:bin 0.500000 0.890000 1.390000 ( 3.118490)
get: libm:ascii 0.820000 1.000000 1.820000 ( 3.914436)
0.47:
set: libm:ascii 0.670000 1.240000 1.910000 ( 4.110841)
set: libm:bin 0.570000 1.340000 1.910000 ( 3.962299)
get: libm:ascii 0.790000 1.240000 2.030000 ( 4.129513)
@evan Do you know where the performance regressions in the new libmemcached are? You mentioned something about added indirection via function pointers.
Not off the top of my head, sorry.
It hasn't been upgraded in a year, and some of the stuff in their changelog looks compelling.