Hello, Thanks for the response and the critiques.
> A couple of comments. > > + if (option_bool(OPT_BIASEDRR)) > + #define BIASED_RESOLUTION] > > looks very strange. The patch does the functionality i had in mind with the least delta. That code is there to essentially mark out the changes that i had done. Looks strange, but does the job and satisfies the compiler xD. I was working on this (hypothetical) usecase: Given that dnsmasq is used on routers,gateways. This provides the end user or other programs on the node,an option at runtime to prefer nameservers in resolv.conf over other upstream nameservers. Then third party public DNS servers could be given to dnsmasq, using the command line option -S which, would have the least priority when the biased round robin option is in effect (that shared memory table is being used judiciously now and manager is a happy man/women) > If the motivation is that different > servers can produce different answers then that is a problem, but it > shouldn't be solved in this way. This was exactly what it was trying to solve. But also provide a way to do it when you are shipping firmware updates and the constraints that come with adopting that work flow. Trying to help, Harish Badrinath
