Looking through the code.. mget() going to be quite slow I suspect because of the memory allocation that will be done because of it.
I modified mget() to not do a malloc, and instead use a vector for the key sizes.
I suspect we could look at the lower level interfaces and find a way of getting the advantages of mget() without doing the vector overhead. Probably just a matter of flipping the bit for pipelining requests and sending down directly the bits from the std:string.
Looking through the code.. mget() going to be quite slow I suspect because of the memory allocation that will be done because of it.
I modified mget() to not do a malloc, and instead use a vector for the key sizes.
I suspect we could look at the lower level interfaces and find a way of getting the advantages of mget() without doing the vector overhead. Probably just a matter of flipping the bit for pipelining requests and sending down directly the bits from the std:string.
Code is now in trunk.