Replacement Reader (2000 Gold)

[mikera]

Currently Convex uses Parboiled (https://github.com/sirthias/parboiled) as a Reader. While parboiled does its job well as a fast PEG Parser, it has a few problems:

• It isn't being actively maintained
• It uses a lot of bytecode magic to build parsers with ASM, which causes JVM compatibility and build issues

Objective of this bounty it to replace Parboiled with a dedicated Convex reader. Requirements:

• Must be able to express Reader rules in a customisable way, similar to Parboiled
• Must produce full CVM values (Cells) as parse results
• Must be pure Java
• Must pass all current Reader tests in Convex main repo
• Must support meaningful parse error messages

[Engelberg]

Can you post some links to the current convex grammar and parsing code? I'm interested in seeing what the existing code looks like to get a feeling for what would be involved in modifying it. Thanks.

[mikera]

It's pretty self-contained in the Reader file: https://github.com/Convex-Dev/convex/blob/master/src/main/java/convex/core/lang/Reader.java

Should be plenty of tests to verify the equivalence of any replacement.

As mentioned in the call, I think ANTLR may be a good choice.

[mikera]

https://github.com/antlr/grammars-v4/blob/master/clojure/Clojure.g4

Clojure grammar example!

[mikera]

@Engelberg are you interested in taking a crack at this? Would be nice to get it in for the Alpha Release

[Engelberg]

I do find the challenge interesting, but I'm deep into a puzzle project right now that is taking all my time. I'm glad to see Bill is making rapid progress on this!

[mikera]

OK thanks for letting me know @Engelberg .

@billbsing I think this is potentially a blocker for alpha so I'm going to take a crack at it myself. Let me know if you want to collab!

[mikera]

Closing because already completed! ANTLR Reader now ready for use.