-
Committer:
John Arbash Meinel
-
Date:
2008-11-07 23:34:46 UTC
-
mto:
(3868.1.1 integration)
-
mto:
This revision was merged to the branch mainline in
revision
3870.
-
Revision ID:
john@arbash-meinel.com-20081107233446-g6h0i9lunaqx7uul
It turns out that we read the pack-names file 3-times because
of inefficiencies in the BTree logic.
1) iter_all_entries() reads the root node, and then proceeds to read all
leaf nodes, but for pack-names, there is only the root node.
2) _read_nodes() reads the whole file when it doesn't know the page size,
but only does so to get the size, and then continues on to read the
requested nodes. Instead, just use what we read.