[chord] merkle

[Benjie Chen]

i am still trying to track down this memory leak, but in the process i
was looking at the merkle code. it seems like now on every block insert
we are enumerating over the entire DB. this seems poor, in terms of
performance.

the fix would be to re-write some of the code so we don't enumerate
frequently. currently we enumerate on every insert to make
sure the merkle tree is fresh, but it seems that we only need to make
sure the tree is fresh when the merkle protocol runs.

any insights?

benjie


-- 
benjie chen
benjie at lcs.mit.edu