Visual Studio Code Extension and Auto-completion for Mizar Language

Abstract

Until now, the editor extension for the Mizar language has been developed as an Emacs plugin. However, since Emacs is an old editor with characteristic key bindings, the number of users has decreased in recent years. For this reason, there is a growing interest in Visual Studio Code as an IDE for interactive theorem provers, and extensions for Lean, Coq, PVS have already been developed. We have developed a new extension for the Mizar language on Visual Studio Code. In this paper, we outline the implementation status of each function and then propose an auto-completion algorithm using n-gram. We measured the reduction rate of keystrokes and the prediction accuracy when 0 to 3 characters are input. As a result, we confirmed that the number of keystrokes can be reduced by 21% and that the correct answer rate within the proposed three candidates exceeds 80% when the user enters two characters. The advantages of our proposed auto-completion algorithm are its ease of implementation and high accuracy.