JavaScript is the de facto programming language of the Web. Not only is it widely used for client-side programming [1], but it is also used for programming server-side applications via Node.js and it even runs on small embedded devices.
To guarantee that JavaScript programs behave consistently across all platforms, the major browser vendors very early agreed to standardise the language, creating the ECMAScript standard [2]. The ECMAScript standard, currently in its 10th edition, is a long, complex document, written in English, which describes the syntax and semantics of JavaScript. It is written as if it were the pseudo-code of a JavaScript interpreter in that it describes each statement of the language operationally, detailing each step of the evaluation.
All the major JavaScript stakeholders (browser vendors, web security companies, governmental agencies, international research groups, etc) accept the ECMAScript standard to be the canonical version of the language and try to conform to it as closely as possible in their own implementations and models. This makes the ECMAScript standard one of the cornerstones of the internet of today. Despite its importance, the ECMAScript standard is in constant change. In fact, the ECMA Committee holds various meetings during the year to discuss modifications and additions to the standard in response to requests from the JavaScript community. These modifications and additions must then be propagated to the implementations of the language, which is often highly non-trivial as these implementations rely on various optimisation strategies in order to be as performant as possible [3,4].
The goal of this project is to design and implement a mechanism to semi-automatically generate an executable version of the standard, the Live JavaScript Specification, from its official textual description. The Live JavaScript Specification will be primarily used as an oracle to test other implementations of the language, thereby contributing to the effort of making JavaScript run uniformly and correctly across all platforms. Further applications can also be envisaged; for instance, one could use the Live JavaScript Specification as a stepping stone for automatically synthesising a correct-by-construction JavaScript compiler [5] or an official language test-suite.
This project will consist of three main tasks. First, the student will design and implement ECMA-SL, a domain-specific language [6] for specifying and running the ECMAScript standard. Then, the student will apply rule-based natural language processing techniques to (semi)-automatically obtain a JavaScript interpreter written in ECMA-SL from the textual description of the standard. Lastly, the student will test the obtained interpreter against Test262, the official JavaScript test suite.
[1] w3techs.com/technologies/details/cp-javascript
[2] ECMAScript Committee. The 10th Edition of the ECMAScript Language Specification. ECMA. 2019.
[3] M. Serrano. JavaScript Ahead of Time Compilation. DLS'19.
[4] T. Rompf et al. Surgical Precision JIT Compilers. PLDI'14.
[5] N. Jones et al. Partial Evaluation and Automatic Program Generation. Prentice Hall. 1999.
[6] M. Mernik et al. When and How to Develop Domain Specific Languages. ACM Computing Surveys. 2005.