Hi Utomo, *, On Sat, Oct 07, 2006 at 12:56:23PM +0700, Utomo wrote: > > Yes it will be difficult to measure. > Or do you think better to start it first (the bounty) > But we will post the detail method to measure it as well the sample files, > in one month ? > (I know it is not so good idea)
I don't think it is a good idea to start and afterwards define the rules. > Any suggestion ? Try to find a way to measure the success. Think of people that claim the bounty. They'll write: "On my system this patch will result in a 20% improvement." Then you apply the patch and try yourself and don't notice any difference. Will you pay? If not, the one providing the patch will be very pissed, because I don't think he lies, it really has an impact on his system or in the way he measures it... If you pay, then the intention you had with it is somewhat nullified. You pay for an improvement "on the paper", but it doesn't feel faster/better for you. It is best to have an automatic testcase that times and measures the effect. I know Sun does performance tests, so maybe somebody from Sun's QA team can tell us how they do the tests. But don't make it to fine grained. Your initial draft was based on the percentage improvement - better make it bigger steps. I'm all for doing bounties, so don't get this wrong. This is just to avoid possible conflicts. ciao Christian -- NP: Corrosion Of Conformity - Shake Like You --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
