minor changes

putting this on hold; implementing TTG first

CHANGELOG.md

@@ -0,0 +1,19 @@
+# unreleased
+
+* New tag syntax:
+  ```hs
+  case x of
+    { 1 -> something
+    ; 2 -> another
+    }
+  ```
+  is now written as
+  ```hs
+  case x of
+    { <1> -> something
+    ; <2> -> another
+    }
+  ```
+
+# Release 1.0.0
+

README.md

@@ -30,12 +30,12 @@ $ rlpc -ddump-opts t.hs
 
 ### Potential Features
 Listed in order of importance.
-- [ ] ADTs
+- [x] ADTs
-- [ ] First-class functions
+- [x] First-class functions
 - [ ] Higher-kinded types
 - [ ] Typeclasses
-- [ ] Parametric polymorphism
+- [x] Parametric polymorphism
-- [ ] Hindley-Milner type inference
+- [x] Hindley-Milner type inference
 - [ ] Newtype coercion
 - [ ] Parallelism
 
@@ -66,32 +66,60 @@ Listed in order of importance.
         - [ ] TCO
         - [ ] DCE
 - [ ] Frontend
-    - [ ] High-level language
+    - [x] High-level language
-        - [ ] AST
+        - [x] AST
-        - [ ] Lexer
+        - [x] Lexer
-        - [ ] Parser
+        - [x] Parser
     - [ ] Translation to the core language
         - [ ] Constraint solver
         - [ ] `do`-notation
     - [x] CLI
 - [ ] Documentation
-    - [ ] State transition rules
+    - [x] State transition rules
     - [ ] How does the evaluation model work?
+    - [ ] The Hindley-Milner type system
     - [ ] CLI usage
     - [ ] Tail call optimisation
-    - [x] Parsing rlp
+    - [ ] Parsing rlp
 - [ ] Tests
     - [x] Generic example programs
     - [ ] Parser
 
-### December Release Plan
+### ~~December Release Plan~~
-- [ ] Tests
+- [x] Tests
     - [ ] Core lexer
     - [ ] Core parser
-    - [ ] Evaluation model
+    - [x] Evaluation model
 - [ ] Benchmarks
-- [ ] Stable Core lexer
+- [x] Stable Core lexer
-- [ ] Stable Core parser
+- [x] Stable Core parser
-- [ ] Stable evaluation model
+- [x] Stable evaluation model
-    - [ ] Garbage Collection
+    - [x] Garbage Collection
 - [ ] Stable documentation for the evaluation model
+
+### January Release Plan
+- [ ] Beta rl' to Core
+- [ ] UX improvements
+    - [ ] Actual compiler errors -- no more unexceptional `error` calls
+    - [ ] Better CLI dump flags
+    - [ ] Annotate the AST with token positions for errors
+- [ ] More examples
+
+### March Release Plan
+- [ ] Tests
+    - [ ] rl' parser
+    - [ ] rl' lexer
+
+### Indefinite Release Plan
+
+This list is more concrete than the milestones, but likely further in the future
+than the other release plans.
+
+- [ ] Stable rl' to Core
+- [ ] Core polish
+    - [ ] Better, stable parser
+    - [ ] Better, stable lexer
+    - [ ] Less hacky handling of named data
+    - [ ] Less hacky pragmas
+- [ ] GM to LLVM
+

app/Main.hs

@@ -63,7 +63,7 @@ options = RLPCOptions
 evaluatorReader :: ReadM Evaluator
 evaluatorReader = maybeReader $ \case
     "gm"  -> Just EvaluatorGM
-    "tim" -> Just EvaluatorTI
+    "ti"  -> Just EvaluatorTI
     _     -> Nothing
 
 mmany :: (Alternative f, Monoid m) => f m -> f m

doc/src/commentary/gm.rst

@@ -112,5 +112,3 @@ The way around this is quite simple: simply offset the stack when w
    :end-before: -- << [ref/compileC]
    :caption: src/GM.hs
 
-
-

doc/src/commentary/layout-lexing.rst

@@ -2,16 +2,21 @@ Lexing, Parsing, and Layouts
 ============================
 
 The C-style languages of my previous experiences have all had quite trivial
-lexical analysis stages, peaking in complexity when I streamed tokens lazily in
+lexical analysis stages: you ignore all whitespace and point out the symbols you
-C. The task of tokenising a C-style language is very simple in description: you
+recognise. If you don't recognise something, check if it's a literal or an
-ignore all whitespace and point out what you recognise. If you don't recognise
+identifier. Should it be neither, return an error.
-something, check if it's a literal or an identifier. Should it be neither,
-return an error.
 
-On paper, both lexing and parsing a Haskell-like language seem to pose a few
+In contrast, both lexing and parsing a Haskell-like language poses a number of
 greater challenges. Listed by ascending intimidation factor, some of the
 potential roadblocks on my mind before making an attempt were:
 
+* Context-sensitive keywords; Haskell allows for some words to be used as
+  identifiers in appropriate contexts, such as :code:`family`, :code:`role`,
+  :code:`as`. Reading a note_ found in `GHC's lexer`_, it appears that keywords
+  are only considered in bodies for which their use is relevant, e.g.
+  :code:`family` and :code:`role` in type declarations, :code:`as` after
+  :code:`case`; :code:`if`, :code:`then`, and :code:`else` in expressions, etc.
+
 * Operators; Haskell has not only user-defined infix operators, but user-defined
   precedence levels and associativities. I recall using an algorithm that looked
   up infix, prefix, postfix, and even mixfix operators up in a global table to
@@ -19,17 +24,9 @@ potential roadblocks on my mind before making an attempt were:
   stored in the table). I never modified the table at runtime, however this
   could be a very nice solution for Haskell.
 
-* Context-sensitive keywords; Haskell allows for some words to be used as identifiers in
-  appropriate contexts, such as :code:`family`, :code:`role`, :code:`as`.
-  Reading a note_ found in `GHC's lexer`_,
-  it appears that keywords are only considered in bodies for which their use is
-  relevant, e.g. :code:`family` and :code:`role` in type declarations,
-  :code:`as` after :code:`case`; :code:`if`, :code:`then`, and :code:`else` in
-  expressions, etc.
-
 * Whitespace sensitivity; While I was comfortable with the idea of a system
-  similar to Python's INDENT/DEDENT tokens, Haskell seemed to use whitespace to
+  similar to Python's INDENT/DEDENT tokens, Haskell's layout system is based on
-  section code in a way that *felt* different.
+  alignment and is very generous with line-folding.
 
 .. _note: https://gitlab.haskell.org/ghc/ghc/-/wikis/commentary/coding-style#2-using-notes
 .. _GHC's lexer: https://gitlab.haskell.org/ghc/ghc/-/blob/master/compiler/GHC/Parser/Lexer.x#L1133
@@ -45,9 +42,9 @@ We will compare and contrast with Python's lexical analysis. Much to my dismay,
 Python uses newlines and indentation to separate statements and resolve scope
 instead of the traditional semicolons and braces found in C-style languages (we
 may generally refer to these C-style languages as *explicitly-sectioned*).
-Internally during tokenisation, when the Python lexer begins a new line, they
+Internally during tokenisation, when the Python lexer encounters a new line, the
-compare the indentation of the new line with that of the previous and apply the
+indentation of the new line is compared with that of the previous and the
-following rules:
+following rules are applied:
 
 1. If the new line has greater indentation than the previous, insert an INDENT
    token and push the new line's indentation level onto the indentation stack
@@ -60,44 +57,37 @@ following rules:
 3. If the indentation is equal, insert a NEWLINE token to terminate the previous
    line, and leave it at that!
 
-Parsing Python with the INDENT, DEDENT, and NEWLINE tokens is identical to
+On the parser's end, the INDENT, DEDENT, and NEWLINE tokens are identical to
-parsing a language with braces and semicolons. This is a solution pretty in line
+braces and semicolons. In developing our *layout* rules, we will follow in the
-with Python's philosophy of the "one correct answer" (TODO: this needs a
+pattern of translating the whitespace-sensitive source language to an explicitly
-source). In developing our *layout* rules, we will follow in the pattern of
+sectioned language.
-translating the whitespace-sensitive source language to an explicitly sectioned
-language.
 
 But What About Haskell?
 ***********************
 
-We saw that Python, the most notable example of an implicitly sectioned
+Parsing Haskell -- and thus rl' -- is only slightly more complex than Python,
-language, is pretty simple to lex. Why then am I so afraid of Haskell's layouts?
+but the design is certainly more sensitive. 
-To be frank, I'm far less scared after asking myself this -- however there are
-certainly some new complexities that Python needn't concern. Haskell has
-implicit line *continuation*: forms written over multiple lines; indentation
-styles often seen in Haskell are somewhat esoteric compared to Python's
-"s/[{};]//".
 
 .. code-block:: haskell
 
-   -- line continuation
+   -- line folds
    something = this is a
        single expression
 
    -- an extremely common style found in haskell
-   data Python = Users
+   data Some = Data
-       { are        :: Crying
+       { is    :: Presented
-       , right      :: About
+       , in    :: This
-       , now        :: Sorry
+       , silly :: Style
        }
 
-   -- another formatting oddity
+   -- another style oddity
    -- note that this is not a single
    -- continued line! `look at`,
-   -- `this`, and `alignment` are all
+   -- `this odd`, and `alignment` are all
-   -- separate expressions!
+   -- discrete items!
    anotherThing = do look at
-                     this
+                     this odd
                      alignment
 
 But enough fear, lets actually think about implementation. Firstly, some
@@ -233,3 +223,4 @@ References
 
 * `Haskell syntax reference
   <https://www.haskell.org/onlinereport/haskell2010/haskellch10.html>`_
+

doc/src/commentary/type-inference.rst

@@ -0,0 +1,5 @@
+Type Inference in rl'
+=====================
+
+rl' implements type inference via the Hindley-Milner type system.
+

doc/src/references/rlp-inference-rules.rst

@@ -0,0 +1,17 @@
+rl' Inference Rules
+===================
+
+.. rubric::
+   [Var]
+
+.. math::
+   \frac{x : \tau \in \Gamma}
+        {\Gamma \vdash x : \tau}
+
+.. rubric::
+   [App]
+
+.. math::
+   \frac{\Gamma \vdash f : \alpha \to \beta \qquad \Gamma \vdash x : \alpha}
+        {\Gamma \vdash f x : \beta}
+

examples/factorial.hs

@@ -1,6 +1,6 @@
 fac n = case (==#) n 0 of
-    { 1 -> 1
+    { <1> -> 1
-    ; 0 -> (*#) n (fac ((-#) n 1))
+    ; <0> -> (*#) n (fac ((-#) n 1))
     };
 
 main = fac 3;

examples/sumList.hs

@@ -2,8 +2,8 @@ nil = Pack{0 0};
 cons x y = Pack{1 2} x y;
 list = cons 1 (cons 2 (cons 3 nil));
 sum l = case l of
-    { 0      -> 0
+    { <0>      -> 0
-    ; 1 x xs -> (+#) x (sum xs)
+    ; <1> x xs -> (+#) x (sum xs)
     };
 main = sum list;
 

programming-language-checklist

@@ -0,0 +1,105 @@
+Programming Language Checklist
+by Colin McMillen, Jason Reed, and Elly Fong-Jones, 2011-10-10.
+
+You appear to be advocating a new:
+[x] functional  [ ] imperative  [ ] object-oriented  [ ] procedural [ ] stack-based
+[ ] "multi-paradigm"  [x] lazy  [ ] eager  [x] statically-typed  [ ] dynamically-typed
+[x] pure  [ ] impure  [ ] non-hygienic  [ ] visual  [x] beginner-friendly
+[ ] non-programmer-friendly  [ ] completely incomprehensible
+programming language.  Your language will not work.  Here is why it will not work.
+
+You appear to believe that:
+[ ] Syntax is what makes programming difficult
+[x] Garbage collection is free                [x] Computers have infinite memory
+[x] Nobody really needs:
+    [x] concurrency  [x] a REPL  [x] debugger support  [x] IDE support  [x] I/O
+    [x] to interact with code not written in your language
+[ ] The entire world speaks 7-bit ASCII
+[ ] Scaling up to large software projects will be easy
+[ ] Convincing programmers to adopt a new language will be easy
+[ ] Convincing programmers to adopt a language-specific IDE will be easy
+[ ] Programmers love writing lots of boilerplate
+[ ] Specifying behaviors as "undefined" means that programmers won't rely on them
+[ ] "Spooky action at a distance" makes programming more fun
+
+Unfortunately, your language (has/lacks):
+[x] comprehensible syntax  [ ] semicolons  [x] significant whitespace  [ ] macros
+[ ] implicit type conversion  [ ] explicit casting  [x] type inference
+[ ] goto  [ ] exceptions  [x] closures  [x] tail recursion  [ ] coroutines
+[ ] reflection  [ ] subtyping  [ ] multiple inheritance  [x] operator overloading
+[x] algebraic datatypes  [x] recursive types  [x] polymorphic types
+[ ] covariant array typing  [x] monads  [ ] dependent types
+[x] infix operators  [x] nested comments  [ ] multi-line strings  [ ] regexes
+[ ] call-by-value  [x] call-by-name  [ ] call-by-reference  [ ] call-cc
+
+The following philosophical objections apply:
+[ ] Programmers should not need to understand category theory to write "Hello, World!"
+[ ] Programmers should not develop RSI from writing "Hello, World!"
+[ ] The most significant program written in your language is its own compiler
+[x] The most significant program written in your language isn't even its own compiler
+[x] No language spec
+[x] "The implementation is the spec"
+   [ ] The implementation is closed-source  [ ] covered by patents  [ ] not owned by you
+[ ] Your type system is unsound  [ ] Your language cannot be unambiguously parsed
+   [ ] a proof of same is attached
+   [ ] invoking this proof crashes the compiler
+[x] The name of your language makes it impossible to find on Google
+[x] Interpreted languages will never be as fast as C
+[ ] Compiled languages will never be "extensible"
+[ ] Writing a compiler that understands English is AI-complete
+[ ] Your language relies on an optimization which has never been shown possible
+[ ] There are less than 100 programmers on Earth smart enough to use your language
+[ ] ____________________________ takes exponential time
+[ ] ____________________________ is known to be undecidable
+
+Your implementation has the following flaws:
+[ ] CPUs do not work that way
+[ ] RAM does not work that way
+[ ] VMs do not work that way
+[ ] Compilers do not work that way
+[ ] Compilers cannot work that way
+[ ] Shift-reduce conflicts in parsing seem to be resolved using rand()
+[ ] You require the compiler to be present at runtime
+[ ] You require the language runtime to be present at compile-time
+[ ] Your compiler errors are completely inscrutable
+[ ] Dangerous behavior is only a warning
+[ ] The compiler crashes if you look at it funny
+[x] The VM crashes if you look at it funny
+[x] You don't seem to understand basic optimization techniques
+[x] You don't seem to understand basic systems programming
+[ ] You don't seem to understand pointers
+[ ] You don't seem to understand functions
+
+Additionally, your marketing has the following problems:
+[x] Unsupported claims of increased productivity
+[x] Unsupported claims of greater "ease of use"
+[ ] Obviously rigged benchmarks
+   [ ] Graphics, simulation, or crypto benchmarks where your code just calls
+       handwritten assembly through your FFI
+   [ ] String-processing benchmarks where you just call PCRE
+   [ ] Matrix-math benchmarks where you just call BLAS
+[x] Noone really believes that your language is faster than:
+    [x] assembly  [x] C  [x] FORTRAN  [x] Java  [x] Ruby  [ ] Prolog
+[ ] Rejection of orthodox programming-language theory without justification
+[x] Rejection of orthodox systems programming without justification
+[ ] Rejection of orthodox algorithmic theory without justification
+[ ] Rejection of basic computer science without justification
+
+Taking the wider ecosystem into account, I would like to note that:
+[x] Your complex sample code would be one line in: examples/
+[ ] We already have an unsafe imperative language
+[ ] We already have a safe imperative OO language
+[x] We already have a safe statically-typed eager functional language
+[ ] You have reinvented Lisp but worse
+[ ] You have reinvented Javascript but worse
+[ ] You have reinvented Java but worse
+[ ] You have reinvented C++ but worse
+[ ] You have reinvented PHP but worse
+[ ] You have reinvented PHP better, but that's still no justification
+[ ] You have reinvented Brainfuck but non-ironically
+
+In conclusion, this is what I think of you:
+[ ] You have some interesting ideas, but this won't fly.
+[x] This is a bad language, and you should feel bad for inventing it.
+[ ] Programming in this language is an adequate punishment for inventing it.
+

rlp.cabal

@@ -7,7 +7,7 @@ license:            GPL-2.0-only
 -- license-file:       LICENSE
 author:             crumbtoo
 maintainer:         crumb@disroot.org
--- copyright:
+copyright:          Madeleine Sydney Ślaga
 category:           Language
 build-type:         Simple
 extra-doc-files:    README.md

src/Compiler/RLPC.hs

@@ -28,14 +28,12 @@ module Compiler.RLPC
     , evalRLPCIO
     , evalRLPC
     , rlpcLogFile
-    , rlpcDebugOpts
+    , rlpcDFlags  
     , rlpcEvaluator
     , rlpcInputFiles
     , DebugFlag(..)
-    , whenFlag
+    , whenDFlag
-    , flagDDumpEval
+    , whenFFlag
-    , flagDDumpOpts
-    , flagDDumpAST
     , def
     , liftErrorful
     )
@@ -43,6 +41,7 @@ module Compiler.RLPC
 ----------------------------------------------------------------------------------
 import Control.Arrow            ((>>>))
 import Control.Exception
+import Control.Monad
 import Control.Monad.Reader
 import Control.Monad.State      (MonadState(state))
 import Control.Monad.Errorful
@@ -51,19 +50,19 @@ import Data.Functor.Identity
 import Data.Default.Class
 import Data.Foldable
 import GHC.Generics             (Generic)
+import Data.Maybe
 import Data.Hashable            (Hashable)
 import Data.HashSet             (HashSet)
 import Data.HashSet             qualified as S
 import Data.Coerce
-import Lens.Micro
+import Lens.Micro.Platform
-import Lens.Micro.TH
 import System.Exit
 ----------------------------------------------------------------------------------
 
 newtype RLPCT m a = RLPCT {
         runRLPCT :: ReaderT RLPCOptions (ErrorfulT (MsgEnvelope RlpcError) m) a
     }
-    deriving (Functor, Applicative, Monad)
+    deriving (Functor, Applicative, Monad, MonadReader RLPCOptions)
 
 type RLPC = RLPCT Identity
 
@@ -98,7 +97,8 @@ liftErrorful e = RLPCT $ lift (fmap liftRlpcError `mapErrorful` e)
 
 data RLPCOptions = RLPCOptions
     { _rlpcLogFile     :: Maybe FilePath
-    , _rlpcDebugOpts   :: DebugOpts
+    , _rlpcDFlags      :: HashSet DebugFlag
+    , _rlpcFFlags      :: HashSet CompilerFlag
     , _rlpcEvaluator   :: Evaluator
     , _rlpcHeapTrigger :: Int
     , _rlpcInputFiles  :: [FilePath]
@@ -113,38 +113,33 @@ data Evaluator = EvaluatorGM | EvaluatorTI
 instance Default RLPCOptions where
     def = RLPCOptions
         { _rlpcLogFile = Nothing
-        , _rlpcDebugOpts = mempty
+        , _rlpcDFlags = mempty
+        , _rlpcFFlags = mempty
         , _rlpcEvaluator = EvaluatorGM
         , _rlpcHeapTrigger = 200
         , _rlpcInputFiles = []
         }
 
-type DebugOpts = HashSet DebugFlag
+-- debug flags are passed with -dFLAG
+type DebugFlag = String
 
-data DebugFlag = DDumpEval
+type CompilerFlag = String
-               | DDumpOpts
-               | DDumpAST
-               deriving (Show, Eq, Generic)
-
-instance Hashable DebugFlag
 
 makeLenses ''RLPCOptions
 pure []
 
-whenFlag :: (MonadReader s m) => SimpleGetter s Bool -> m () -> m ()
+-- TODO: rewrite this with prisms once microlens-pro drops :3
-whenFlag l m = asks (^. l) >>= \a -> if a then m else pure ()
+whenDFlag :: (Monad m) => DebugFlag -> RLPCT m () -> RLPCT m ()
+whenDFlag f m = do
+    -- mfw no `At` instance for HashSet
+    fs <- view rlpcDFlags
+    let a = S.member f fs
+    when a m
 
--- there's probably a better way to write this. my current knowledge of lenses
+whenFFlag :: (Monad m) => CompilerFlag -> RLPCT m () -> RLPCT m ()
--- is too weak.
+whenFFlag f m = do
-flagGetter :: DebugFlag -> SimpleGetter RLPCOptions Bool
+    -- mfw no `At` instance for HashSet
-flagGetter d = to $ \s -> s ^. rlpcDebugOpts & S.member d
+    fs <- view rlpcFFlags
-
+    let a = S.member f fs
-flagDDumpEval :: SimpleGetter RLPCOptions Bool
+    when a m
-flagDDumpEval = flagGetter DDumpEval
-
-flagDDumpOpts :: SimpleGetter RLPCOptions Bool
-flagDDumpOpts = flagGetter DDumpOpts
-
-flagDDumpAST :: SimpleGetter RLPCOptions Bool
-flagDDumpAST = flagGetter DDumpAST
 

src/Core/Examples.hs

@@ -4,12 +4,7 @@ Description : Core examples (may eventually be unit tests)
 -}
 {-# LANGUAGE QuasiQuotes #-}
 {-# LANGUAGE OverloadedStrings #-}
-module Core.Examples
+module Core.Examples where
-    ( fac3
-    , sumList
-    , constDivZero
-    , idCase
-    ) where
 ----------------------------------------------------------------------------------
 import Core.Syntax
 import Core.TH
@@ -147,8 +142,8 @@ simple1 = [coreProg|
 caseBool1 :: Program'
 caseBool1 = [coreProg|
     _if c x y = case c of
-        { 1 -> x
+        { <1> -> x
-        ; 0 -> y
+        ; <0> -> y
         };
 
     false = Pack{0 0};
@@ -160,8 +155,8 @@ caseBool1 = [coreProg|
 fac3 :: Program'
 fac3 = [coreProg|
     fac n = case (==#) n 0 of
-        { 1 -> 1
+        { <1> -> 1
-        ; 0 -> (*#) n (fac ((-#) n 1))
+        ; <0> -> (*#) n (fac ((-#) n 1))
         };
 
     main = fac 3;
@@ -175,8 +170,8 @@ sumList = [coreProg|
         cons x y = Pack{1 2} x y;
         list = cons 1 (cons 2 (cons 3 nil));
         sum l = case l of
-            { 0      -> 0
+            { <0>      -> 0
-            ; 1 x xs -> (+#) x (sum xs)
+            ; <1> x xs -> (+#) x (sum xs)
             };
         main = sum list;
     |]
@@ -192,10 +187,36 @@ idCase = [coreProg|
         id x = x;
 
         main = id (case Pack{1 0} of
-            { 1 -> (+#) 2 3
+            { <1> -> (+#) 2 3
             })
     |]
 
+-- NOTE: the GM primitive (==#) returns an untyped constructor with tag 1 for
+-- true, and 0 for false. See: GM.boxBool
+namedBoolCase :: Program'
+namedBoolCase = [coreProg|
+        {-# PackData True 1 0 #-}
+        {-# PackData False 0 0 #-}
+        main = case (==#) 1 1 of
+            { True -> 123
+            ; False -> 456
+            }
+    |]
+
+namedConsCase :: Program'
+namedConsCase = [coreProg|
+        {-# PackData Nil  0 0 #-}
+        {-# PackData Cons 1 2 #-}
+        Nil = Pack{0 0};
+        Cons = Pack{1 2};
+        foldr f z l = case l of
+            { Nil       -> z
+            ; Cons x xs -> f x (foldr f z xs)
+            };
+        list = Cons 1 (Cons 2 (Cons 3 Nil));
+        main = foldr (+#) 0 list
+    |]
+
 -- corePrelude :: Module Name
 -- corePrelude = Module (Just ("Prelude", [])) $
 --     -- non-primitive defs

src/Core/HindleyMilner.hs

@@ -55,11 +55,14 @@ instance IsRlpcError TypeError where
     liftRlpcError = \case
         -- todo: use anti-parser instead of show
         TyErrCouldNotUnify t u -> Text
-            [ T.pack $ printf "Could not match type `%s' with `%s'."
+            [ T.pack $ printf "Could not match type `%s` with `%s`."
                               (show t) (show u)
             , "Expected: " <> tshow t
             , "Got: " <> tshow u
             ]
+        TyErrUntypedVariable n -> Text
+            [ "Untyped (likely undefined) variable `" <> n <> "`"
+            ]
         TyErrRecursiveType t x -> Text
             [ T.pack $ printf "recursive type error lol"
             ]
@@ -157,7 +160,12 @@ gather = \g e -> runStateT (go g e) ([],0) <&> \ (t,(cs,_)) -> (t,cs) where
         Let Rec bs e -> do
             g' <- buildLetrecContext g bs
             go g' e
-
+        Lam bs e -> case bs of
+            [x] -> do
+                tx <- uniqueVar
+                let g' = (x,tx) : g
+                te <- go g' e
+                pure (tx :-> te)
         -- TODO lambda, case
 
     buildLetrecContext :: Context' -> [Binding']

src/Core/Lex.x

@@ -65,6 +65,8 @@ $white_no_nl = $white # $nl
 
 @decimal = $digit+
 
+@alttag    = "<" $digit+ ">"
+
 rlp :-
 
 <0>
@@ -92,6 +94,8 @@ rlp :-
     "="                     { constTok TokenEquals }
     "->"                    { constTok TokenArrow }
 
+    @alttag                 { lexWith ( TokenAltTag . read @Int . T.unpack
+                                      . T.drop 1 . T.init ) }
     @varname                { lexWith TokenVarName }
     @conname                { lexWith TokenConName }
     @varsym                 { lexWith TokenVarSym }
@@ -135,6 +139,7 @@ data CoreToken = TokenLet
                | TokenConName Name
                | TokenVarSym Name
                | TokenConSym Name
+               | TokenAltTag Tag
                | TokenEquals
                | TokenLParen
                | TokenRParen

src/Core/Parse.y

@@ -3,7 +3,7 @@
 Module : Core.Parse
 Description : Parser for the Core language
 -}
-{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE OverloadedStrings, ViewPatterns #-}
 module Core.Parse
     ( parseCore
     , parseCoreExpr
@@ -23,7 +23,9 @@ import Compiler.RLPC
 import Lens.Micro
 import Data.Default.Class   (def)
 import Data.Hashable        (Hashable)
+import Data.List.Extra
 import Data.Text.IO         qualified as TIO
+import Data.Text            (Text)
 import Data.Text            qualified as T
 import Data.HashMap.Strict  qualified as H
 }
@@ -49,6 +51,7 @@ import Data.HashMap.Strict  qualified as H
       varsym          { Located _ _ _ (TokenVarSym $$) }
       conname         { Located _ _ _ (TokenConName $$) }
       consym          { Located _ _ _ (TokenConSym $$) }
+      alttag          { Located _ _ _ (TokenAltTag $$) }
       word            { Located _ _ _ (TokenWord $$) }
       'λ'             { Located _ _ _ TokenLambda }
       '->'            { Located _ _ _ TokenArrow }
@@ -82,6 +85,15 @@ Program         : ScTypeSig ';' Program         { insTypeSig $1 $3 }
                 | ScTypeSig OptSemi             { singletonTypeSig $1 }
                 | ScDef     ';' Program         { insScDef $1 $3 }
                 | ScDef     OptSemi             { singletonScDef $1 }
+                | TLPragma  Program             {% doTLPragma $1 $2 }
+                | TLPragma                      {% doTLPragma $1 mempty }
+
+TLPragma        :: { Pragma }
+                : '{-#' Words '#-}'             { Pragma $2 }
+
+Words           :: { [Text] }
+                : Words word                    { $1 `snoc` $2 }
+                | word                          { [$1] }
 
 OptSemi         :: { () }
 OptSemi         : ';'                           { () }
@@ -94,7 +106,6 @@ ScDefs          :: { [ScDef Name] }
 ScDefs          : ScDef ';' ScDefs              { $1 : $3 }
                 | ScDef ';'                     { [$1] }
                 | ScDef                         { [$1] }
-                | {- epsilon -}                 { [] }
 
 ScDef           :: { ScDef Name }
 ScDef           : Var ParList '=' Expr          { ScDef $1 $2 $4 }
@@ -149,22 +160,15 @@ Alters          : Alter ';' Alters              { $1 : $3 }
                 | Alter                         { [$1] }
 
 Alter           :: { Alter Name }
-Alter           : litint ParList '->' Expr      { Alter (AltData $1) $2 $4 }
+Alter           : alttag ParList '->' Expr      { Alter (AltTag  $1) $2 $4 }
+                | Con    ParList '->' Expr      { Alter (AltData $1) $2 $4 }
 
 Expr1           :: { Expr Name }
 Expr1           : litint                        { Lit $ IntL $1 }
                 | Id                            { Var $1 }
                 | PackCon                       { $1 }
-                | ExprPragma                    { $1 }
                 | '(' Expr ')'                  { $2 }
 
-ExprPragma      :: { Expr Name }
-ExprPragma      : '{-#' Words '#-}'      {% exprPragma $2 }
-
-Words           :: { [String] }
-Words           : word Words                    { T.unpack $1 : $2 }
-                | word                          { [T.unpack $1] }
-
 PackCon         :: { Expr Name }
 PackCon         : pack '{' litint litint '}'    { Con $3 $4 }
 
@@ -229,5 +233,17 @@ happyBind m k = m >>= k
 happyPure :: a -> RLPC a
 happyPure a = pure a
 
+doTLPragma :: Pragma -> Program' -> RLPC Program'
+-- TODO: warn unrecognised pragma
+doTLPragma (Pragma []) p = pure p
+
+doTLPragma (Pragma pr) p = case pr of
+    -- TODO: warn on overwrite
+    ["PackData", n, readt -> t, readt -> a] ->
+        pure $ p & programDataTags . at n ?~ (t,a)
+
+readt :: (Read a) => Text -> a
+readt = read . T.unpack
+
 }
 

src/Core/Syntax.hs

@@ -5,8 +5,14 @@ Description : Core ASTs and the like
 {-# LANGUAGE PatternSynonyms, OverloadedStrings #-}
 {-# LANGUAGE FunctionalDependencies #-}
 {-# LANGUAGE TemplateHaskell #-}
+{-# LANGUAGE DerivingStrategies, DerivingVia #-}
+-- for recursion-schemes
+{-# LANGUAGE DeriveFunctor, DeriveFoldable, DeriveTraversable
+  , TemplateHaskell, TypeFamilies #-}
 module Core.Syntax
     ( Expr(..)
+    , ExprF(..)
+    , ExprF'(..)
     , Type(..)
     , pattern TyInt
     , Lit(..)
@@ -24,9 +30,11 @@ module Core.Syntax
     , Module(..)
     , Program(..)
     , Program'
+    , Pragma(..)
     , unliftScDef
     , programScDefs
     , programTypeSigs
+    , programDataTags
     , Expr'
     , ScDef'
     , Alter'
@@ -40,11 +48,15 @@ import Data.Coerce
 import Data.Pretty
 import Data.List                    (intersperse)
 import Data.Function                ((&))
+import Data.Functor.Foldable
+import Data.Functor.Foldable.TH     (makeBaseFunctor)
 import Data.String
+import Data.HashMap.Strict          (HashMap)
 import Data.HashMap.Strict          qualified as H
 import Data.Hashable
 import Data.Text                    qualified as T
 import Data.Char
+import GHC.Generics
 -- Lift instances for the Core quasiquoters
 import Language.Haskell.TH.Syntax   (Lift)
 import Lens.Micro.TH                (makeLenses)
@@ -99,11 +111,14 @@ data Alter b = Alter AltCon [b] (Expr b)
 
 deriving instance (Eq b) => Eq (Alter b)
 
+newtype Pragma = Pragma [T.Text]
+
 data Rec = Rec
          | NonRec
          deriving (Show, Read, Eq, Lift)
 
-data AltCon = AltData Tag
+data AltCon = AltData Name
+            | AltTag Tag
             | AltLit Lit
             | Default
             deriving (Show, Read, Eq, Lift)
@@ -125,13 +140,20 @@ data Module b = Module (Maybe (Name, [Name])) (Program b)
 
 data Program b = Program
     { _programScDefs   :: [ScDef b]
-    , _programTypeSigs :: H.HashMap b Type
+    , _programTypeSigs :: HashMap b Type
+    -- map constructors to their tag and arity
+    , _programDataTags :: HashMap b (Tag, Int)
     }
-    deriving (Show, Lift)
+    deriving (Show, Lift, Generic)
+    deriving (Semigroup, Monoid)
+        via Generically (Program b)
 
 makeLenses ''Program
+makeBaseFunctor ''Expr
 pure []
 
+type ExprF' = ExprF Name
+
 type Program' = Program Name
 type Expr' = Expr Name
 type ScDef' = ScDef Name
@@ -148,12 +170,6 @@ instance IsString Type where
         | otherwise     = TyVar . fromString $ s
         where (c:_) = s
 
-instance (Hashable b) => Semigroup (Program b) where
-    (<>) = undefined
-
-instance (Hashable b) => Monoid (Program b) where
-    mempty = Program mempty mempty
-
 ----------------------------------------------------------------------------------
 
 class HasRHS s t a b | s -> a, t -> b, s b -> t, t a -> s where
@@ -187,3 +203,8 @@ instance HasLHS (ScDef b) (ScDef b) (b, [b]) (b, [b]) where
         (\ (ScDef n as _) -> (n,as))
         (\ (ScDef _ _ e) (n',as') -> (ScDef n' as' e))
 
+instance HasLHS (Binding b) (Binding b) b b where
+    _lhs = lens
+        (\ (k := _) -> k)
+        (\ (_ := e) k' -> k' := e)
+

src/Core/Utils.hs

@@ -1,16 +1,10 @@
--- for recursion schemes
-{-# LANGUAGE DeriveFunctor, DeriveFoldable, DeriveTraversable #-}
--- for recursion schemes
-{-# LANGUAGE TemplateHaskell, TypeFamilies #-}
-
 module Core.Utils
-    ( bindersOf
+    ( programRhss
-    , rhssOf
+    , programGlobals
     , isAtomic
     -- , insertModule
     , extractProgram
     , freeVariables
-    , ExprF(..)
     )
     where
 ----------------------------------------------------------------------------------
@@ -23,13 +17,11 @@ import Lens.Micro
 import GHC.Exts                     (IsList(..))
 ----------------------------------------------------------------------------------
 
-bindersOf :: (IsList l, Item l ~ b) => [Binding b] -> l
+programGlobals :: Traversal' (Program b) b
-bindersOf bs = fromList $ fmap f bs
+programGlobals = programScDefs . each . _lhs . _1
-    where f (k := _) = k
 
-rhssOf :: (IsList l, Item l ~ Expr b) => [Binding b] -> l
+programRhss :: Traversal' (Program b) (Expr b)
-rhssOf = fromList . fmap f
+programRhss = programScDefs . each . _rhs
-    where f (_ := v) = v
 
 isAtomic :: Expr b -> Bool
 isAtomic (Var _)  = True
@@ -47,8 +39,6 @@ extractProgram (Module _ p) = p
 
 ----------------------------------------------------------------------------------
 
-makeBaseFunctor ''Expr
-
 freeVariables :: Expr' -> Set Name
 freeVariables = cata go
     where
@@ -57,8 +47,8 @@ freeVariables = cata go
         -- TODO: collect free vars in rhss of bs
         go (LetF _ bs e)    = (e `S.union` esFree) `S.difference` ns
             where
-                es = rhssOf bs :: [Expr']
+                es = bs ^.. each . _rhs :: [Expr']
-                ns = bindersOf bs
+                ns = S.fromList $ bs ^.. each . _lhs
                 -- TODO: this feels a little wrong. maybe a different scheme is
                 -- appropriate
                 esFree = foldMap id $ freeVariables <$> es

src/Core2Core.hs

@@ -1,3 +1,4 @@
+{-# LANGUAGE ImplicitParams #-}
 {-# LANGUAGE LambdaCase #-}
 module Core2Core
     ( core2core
@@ -15,11 +16,12 @@ import Data.Set                 (Set)
 import Data.Set                 qualified as S
 import Data.List
 import Control.Monad.Writer
-import Control.Monad.State
+import Control.Monad.State.Lazy
 import Control.Arrow            ((>>>))
 import Data.Text                qualified as T
+import Data.HashMap.Strict      (HashMap)
 import Numeric                  (showHex)
-import Lens.Micro
+import Lens.Micro.Platform
 import Core.Syntax
 import Core.Utils
 ----------------------------------------------------------------------------------
@@ -28,22 +30,35 @@ core2core :: Program' -> Program'
 core2core p = undefined
 
 gmPrep :: Program' -> Program'
-gmPrep p = p' & programScDefs %~ (<>caseScs)
+gmPrep p = p & appFloater (floatNonStrictCases globals)
+             & tagData
     where
-        rhss :: Applicative f => (Expr z -> f (Expr z)) -> Program z -> f (Program z)
-        rhss = programScDefs . each . _rhs
         globals = p ^.. programScDefs . each . _lhs . _1
                 & S.fromList
 
-        -- i kinda don't like that we're calling floatNonStrictCases twice tbh
+tagData :: Program' -> Program'
-        p'      = p & rhss %~ fst . runFloater . floatNonStrictCases globals
+tagData p = let ?dt = p ^. programDataTags
-        caseScs = (p ^.. rhss)
+            in p & programRhss %~ cata go where
-              <&> snd . runFloater . floatNonStrictCases globals
+    go :: (?dt :: HashMap Name (Tag, Int)) => ExprF' Expr' -> Expr'
-                & mconcat
+    go (CaseF  e as)    = Case e (tagAlts <$> as)
+    go x                = embed x
+
+    tagAlts :: (?dt :: HashMap Name (Tag, Int)) => Alter' -> Alter'
+    tagAlts (Alter (AltData c) bs e) = Alter (AltTag tag) bs e
+        where tag = case ?dt ^. at c of
+                Just (t,_) -> t
+                -- TODO: errorful
+                Nothing    -> error $ "unknown constructor " <> show c
+    tagAlts x = x
 
 -- | Auxilary type used in @floatNonSrictCases@
 type Floater = StateT [Name] (Writer [ScDef'])
 
+appFloater :: (Expr' -> Floater Expr') -> Program' -> Program'
+appFloater fl p = p & traverseOf programRhss fl
+                    & runFloater
+                    & \ (me,floats) -> me & programScDefs %~ (<>floats)
+
 runFloater :: Floater a -> (a, [ScDef'])
 runFloater = flip evalStateT ns >>> runWriter
     where

src/GM.hs

@@ -661,7 +661,8 @@ buildInitialHeap (view programScDefs -> ss) = mapAccumL allocateSc mempty compil
         compileC _ (Con t n) = [PushConstr t n]
 
         compileC _ (Case _ _) =
-            error "case expressions may not appear in non-strict contexts :/"
+            error "GM compiler found a non-strict case expression, which should\
+                  \ have been floated by Core2Core.gmPrep. This is a bug!"
 
         compileC _ _ = error "yet to be implemented!"
 
@@ -724,12 +725,16 @@ buildInitialHeap (view programScDefs -> ss) = mapAccumL allocateSc mempty compil
         compileD g as = fmap (compileA g) as
 
         compileA :: Env -> Alter' -> (Tag, Code)
-        compileA g (Alter (AltData t) as e) = (t, [Split n] <> c <> [Slide n])
+        compileA g (Alter (AltTag t) as e) = (t, [Split n] <> c <> [Slide n])
             where
                 n = length as
                 binds = (NameKey <$> as) `zip` [0..]
                 g' = binds ++ argOffset n g
                 c = compileE g' e
+        compileA _ (Alter _ as e) = error "GM.compileA found an untagged\
+                                          \ constructor, which should have\
+                                          \ been handled by Core2Core.gmPrep.\
+                                          \ This is a bug!"
 
         inlineOp1 :: Env -> Instr -> Expr' -> Code
         inlineOp1 g i a = compileE g a <> [i]

tst/Core/HindleyMilnerSpec.hs

@@ -38,6 +38,18 @@ spec = do
         let e = [coreExpr|3|]
         in check' [] (TyCon "Bool") e `shouldSatisfy` isLeft
 
+    it "should infer `fix ((+#) 1)` :: Int" $
+        let g = [ ("fix", ("a" :-> "a") :-> "a")
+                , ("+#", TyInt :-> TyInt :-> TyInt) ]
+            e = [coreExpr|fix ((+#) 1)|]
+        in infer' g e `shouldBe` Right TyInt
+
+    it "should infer mutually recursively defined lists" $
+        let g = [ ("cons", TyInt :-> TyCon "IntList" :-> TyCon "IntList") ]
+            e :: Expr'
+            e = [coreExpr|letrec { as = cons 1 bs; bs = cons 2 as } in as|]
+        in infer' g e `shouldBe` Right (TyCon "IntList")
+
 infer' :: Context' -> Expr' -> Either [TypeError] Type
 infer' g e = case runErrorful $ infer g e of
     (Just t,   _) -> Right t

tst/GMSpec.hs

@@ -27,15 +27,22 @@ spec = do
             in coreRes `shouldBe` arithRes
 
     describe "test programs" $ do
-        it "fac 3" $ do
+        it "fac 3" $
             resultOf Ex.fac3 `shouldBe` Just (NNum 6)
 
-        it "sum [1,2,3]" $ do
+        it "sum [1,2,3]" $
             resultOf Ex.sumList `shouldBe` Just (NNum 6)
 
-        it "k 3 ((/#) 1 0)" $ do
+        it "k 3 ((/#) 1 0)" $
             resultOf Ex.constDivZero `shouldBe` Just (NNum 3)
 
-        it "id (case ... of { ... })" $ do
+        it "id (case ... of { ... })" $
             resultOf Ex.idCase `shouldBe` Just (NNum 5)
 
+        it "bool pattern matching with named constructors" $
+            resultOf Ex.namedBoolCase `shouldBe` Just (NNum 123)
+
+        it "list pattern matching with named constructors" $
+            resultOf Ex.namedConsCase `shouldBe` Just (NNum 6)
+
+