instead of allocating all the data into one big buffer, keep a small input buffer around and decompress directly into a client's read buffer.. r=dougt, sr=darin git-svn-id: svn://10.0.0.236/trunk@139458 18797224-902f-48f8-a5cc-f745e15eee43
instead of allocating all the data into one big buffer, keep a small input buffer around and decompress directly into a client's read buffer.. r=dougt, sr=darin git-svn-id: svn://10.0.0.236/trunk@139458 18797224-902f-48f8-a5cc-f745e15eee43