Analysis of Calling Context Encoding and Decoding Algorithms

Abstract

The calling context of a program is recorded via a call stack for event logging, debugging, and profiling. There are several calling context encoding and decoding schemes that record the calling context of a program. One such scheme we are introducing is DCCE, Distinguished Calling Context Encoding; it can encode a program's calling context using a single integer ID without the need to decode it later. Without the need to decode, DCCE has less overhead costs than other popular encoding schemes. Another advantage of DCCE is that it can distinguish between different calling contexts that have the same encoded ID and different ending nodes/functions. We want to compare DCCE with other existing algorithms in terms of running time and measure the improved efficiency overall. This research paper discusses the practical uses of calling context encoding, implementation methods for DCCE, and the efficiency improvements of DCCE compared to CCTLib encoding. Through our experiment, DCCE outperformed CCTLib by over 2 times of overall execution time.