GoKawiilHi everyone!
I have been working on a new library for writing JVM bytecode with Haskell in a nice, high level way and I’d love some feedback on it! The motivation here is for compilers to the JVM so they can focus on the actual code generation, meanwhile H2JVM takes care of all the messy details like StackMapTable analysis, label/offset resolution, etc.
Here is a quick example taken from the readme. It generates a simple class file with a single method static int add(int, int) which adds 2 numbers:
main :: IO () main = do -- Define the class name, method descriptor, and access flags let className = "Calculator" methodDesc = MethodDescriptor -- int (int, int) [PrimitiveFieldType JInt, PrimitiveFieldType JInt] (TypeReturn (PrimitiveFieldType JInt)) -- Construct the class using ClassBuilder result <- runPureEff $ runErrorNoCallStack @StackMapError $ runClassBuilder className java8 $ do addAccessFlag CPublic -- add the method, which automatically handles stack map analysis, max stack/locals, etc addMethodWithCode "add" [MPublic, MStatic] methodDesc $ do emit $ ILoad 0 emit $ ILoad 1 emit IAdd emit IReturn case result of Left err -> putStrLn $ "Error building class: " <> show err Right (classFile, _) -> do -- Serialise the class to a file let path = classFilePath classFile -- returns "Calculator.class" case classFileBytes classFile of Left err -> putStrLn $ "Error generating bytecode: " <> show err Right bytes -> LBS.writeFile path bytes
This is a less contrived example from real usage of the library in my compiler, which shows how label resolution “just works”. This implements the > operator in my language in the way you’d probably expect
IR.BinaryOp op lhs rhs -> do emitExpr lhs emitExpr rhs case op of IR.GreaterThan -> do trueLabel <- newLabel endLabel <- newLabel emit $ JVM.IfICmp (IfGt trueLabel) -- if_icmpgt jump to trueLabel -- false case emit JVM.IConst0 -- push 0 onto stack emit $ Goto endLabel -- jump to end -- true case emit $ JVM.Label trueLabel emit JVM.IConst1 -- push 1 onto stack emit $ JVM.Label endLabel -- jump to end
The generated code here looks something like this:
29: blah (pushing lhs and rhs onto stack) 32: if_icmpgt 39 -- trueLabel resolved as offset 39 35: iconst_0 36: goto 40 -- endLabel resolved as offset 40 39: iconst_1 40: blah blah (code after the binaryop, whatever that may be)
The library is still in very very very early stages (only a very small subset of the instructions and attributes are supported), but I would love some preliminary feedback on things like the design!
Happy to answer any questions anyone might have too
... continue reading