- Description about the talk
- In parse.y, there is a variable that represents the state of the Lexer, based on the idea that the Parser and the Lexer can be separated. However, in reality, the state of Parser and Lexer are shared and cannot be said to be separated. Also, the Lexer state is manually managed, but historical history has clouded the view, and additions and modifications must be done with care.
With the replacement of the parser generator from Bison to Lrama, the time has come to attack parse.y from the parser generator side, and Lrama is trying to solve this problem by generating an algorithmic parser called PSLR. As a first step, I will show how Lrama can generate a parser for a new algorithm called IELR, which is a prerequisite for PSLR.
IELR is an improved version of LALR and can parse grammars that LALR cannot. In this presentation, I will explain the implementation of Lrama and how the parser is actually generated.