rb: Buffer the slice!s in MemoryBuffer for a significant performance increase [THRIFT-63]

Author: Kevin Ballard <kevin@rapleaf.com>

Currently it buffers up to 4kB before throwing away the data.
Tests with 1MB shows the exact same performance characteristics.


git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@675049 13f79535-47bb-0310-9956-ffa450edef68
2 files changed