Link to full announcement: Compiler Tooling Task Force announcement 8 january 2022 - Google Docs
I know many of us (myself including) think that parsing has historically received too much attention , but you might be interested in @int-index, @knothed, and my ongoing attempts to refactor happy
into libraries so it is a good foundation for all sorts of experiments.
At least I am hoping that demonstrating more recent Haskell development practice with a classic, “used in production” compiler will be a good example / case-study all around.
Alex and happy are in dire need for some attention to stay on top.
Currently you have to pay for speed with UX.
Compared to megaparsec
+ parser-combinators
+ your own packages:
Pros:
- good performance of generated code
- somewhat formal specs
Cons:
- no modularity and reuse
- abysmal error reporting
- illegible generated code riddled with warnings
- language options are restricted by model or being too cumbersome
- entry barrier is higher
There are some heavyweight tools like BNFC, but they’re a whole universe upon themselves.
We certainly are interested in all of those things. The current refactoring phase is to lay a proper foundation for addressing them.