On Sun, Jun 19, 2011 at 11:18:05AM -0400, James E Keenan wrote: > Have we developed those ways of benchmarking the performance of HLLs > built on top of our monthly releases?
I've created the "rpbench" suite for testing many different versions of Rakudo (master) against various versions of Parrot. The code is at http://github.com/pmichaud/rpbench, with published results appearing at http://github.com/pmichaud/rpbench-results . This Friday/Saturday I'll be publishing a new set of results that include the June releases of Parrot and Rakudo. (It takes about 24 hours of dedicated computing time to produce all of the benchmark results.) > 3. We recognized the need for Parrot and Rakudo to have relationship > managers who would "... be each project's principal contact point > for resolution of problems with regard to breakage or deterioration > of performance." I believe we spoke of cotto and whiteknight > filling those roles from the Parrot side, but I don't know whether > we formalized it. And my impression from backscrolling through > #phasers is that Rakudo hasn't made formal assignments either. Our formal assignments will be Moritz Lenz and Patrick Michaud. The latest draft of the relationship manager policy (which is the same as the earlier draft but with the names added) is still at http://pmichaud.com/sandbox/relman-draft.txt . Unless I hear comments or changes that indicate otherwise, Rakudo will consider the policy "adopted" as of this Friday (and will place the text into its docs/ folder then.) Pm