PERFORMANCE EVALUATION FRAMEWORK OF ALL CLASSES OF SELECTORS FOR JAVASCRIPT LIBRARIES

Many JavaScript libraries (JL) have become available in order to facilitate programming of rich clientside interactions in web applications. In some cases these interactions may cause non-negligible overhead on the response time of web applications, esp. on average mobile devices, that negatively influence user experience. An important fraction of this overhead is caused by some JL’s selectors on DOM elements. The aim of this work is to thoroughly study and evaluate the performance of JL’s selectors and categorize them in various performance classes. Towards this purpose, we produced a testsuite of 263 selectors to cover all possible kinds of selectors found in real world cases and evaluate the performance of these selectors on 6 popular JLs, with an enhanced version of slickspeed suite and a large set of jsperf tests. The main result of this work is to introduce a framework (methodology and a set of tools), that can be used by developers to estimate the performance overhead caused by JL DOM selectors in a given web page, discover bad performing selectors and optimize them.