From: Luís O. <lu...@gm...> - 2018-01-08 11:22:40
|
On Mon, Jan 8, 2018 at 9:38 AM Jim Newton <jn...@lr...> wrote: > Hi have a hefty algorithm which I would like to use this weak table feature. > I’m happy to give feedback whether the fix changes anything in my performance. > Can someone explain to me what need to do to test with and without this feature. At first, I misunderstood Douglas's statement and thought that this had been fixed in the (more distant) past, hence my remark about the FIXME. This seems to be the commit where Douglas has improved the garbage collection of weak hash-tables: <https://sourceforge.net/p/sbcl/sbcl/ci/05726638c109c7c3b21c2c282451ac25a3e52538/>. If you'd like to measure the impact on your code, checkout that commit, compile SBCL, measure, do the same with the previous commit and compare. Cheers, Luís |