Add labels to all chapters.

[matthijs/master-project/report.git] / Chapters / Normalization.tex

\chapter[chap:normalization]{Normalization}

% A helper to print a single example in the half the page width. The example
% text should be in a buffer whose name is given in an argument.
%
% The align=right option really does left-alignment, but without the program
% will end up on a single line. The strut=no option prevents a bunch of empty
% space at the start of the frame.
\define[1]\example{
  \framed[offset=1mm,align=right,strut=no]{
    \setuptyping[option=LAM,style=sans,before=,after=]
    \typebuffer[#1]
    \setuptyping[option=none,style=\tttf]
  }
}


% A transformation example
\definefloat[example][examples]
\setupcaption[example][location=top] % Put captions on top

\define[3]\transexample{
  \placeexample[here]{#1}
  \startcombination[2*1]
    {\example{#2}}{Original program}
    {\example{#3}}{Transformed program}
  \stopcombination
}
%
%\define[3]\transexampleh{
%%  \placeexample[here]{#1}
%%  \startcombination[1*2]
%%    {\example{#2}}{Original program}
%%    {\example{#3}}{Transformed program}
%%  \stopcombination
%}

The first step in the core to VHDL translation process, is normalization. We
aim to bring the core description into a simpler form, which we can
subsequently translate into VHDL easily. This normal form is needed because
the full core language is more expressive than VHDL in some areas and because
core can describe expressions that do not have a direct hardware
interpretation.

TODO: Describe core properties not supported in VHDL, and describe how the
VHDL we want to generate should look like.

\section{Goal}
The transformations described here have a well-defined goal: To bring the
program in a well-defined form that is directly translatable to hardware,
while fully preserving the semantics of the program.

This {\em normal form} is again a Core program, but with a very specific
structure. A function in normal form has nested lambda's at the top, which
produce a number of nested let expressions. These let expressions binds a
number of simple expressions in the function and produces a simple identifier.
Every bound value in the let expression is either a simple function
application, a case expression to extract a single element from a tuple
returned by a function or a case expression to choose between two signals
based on some other signal.

This structure is easy to translate to VHDL, since each top level lambda will
be an input port, every bound value will become a concurrent statement (such
as a component instantiation or conditional signal assignment) and the result
variable will become the output port.

An example of a program in canonical form would be:

\startlambda
  -- All arguments are an inital lambda
  λa.λd.λsp.
  -- There are nested let expressions at top level
  let
    -- Unpack the state by coercion
    s = sp :: (Word, Word)
    -- Extract both registers from the state
    r1 = case s of (fst, snd) -> fst
    r2 = case s of (fst, snd) -> snd
    -- Calling some other user-defined function.
    d' = foo d
    -- Conditional connections
    out = case a of
      High -> r1
      Low -> r2
    r1' = case a of
      High -> d
      Low -> r1
    r2' = case a of
      High -> r2
      Low -> d
    -- Packing a tuple
    s' = (,) r1' r2'
    -- Packing the state by coercion
    sp' = s' :: State (Word, Word)
    -- Pack our return value
    res = (,) sp' out
  in
    -- The actual result
    res
\stoplambda

\subsection{Definitions}
In the following sections, we will be using a number of functions and
notations, which we will define here.

\subsubsection{Transformations}
The most important notation is the one for transformation, which looks like
the following:

\starttrans
context conditions
~
from
------------------------            expression conditions
to
~
context additions
\stoptrans

Here, we describe a transformation. The most import parts are \lam{from} and
\lam{to}, which describe the Core expresssion that should be matched and the
expression that it should be replaced with. This matching can occur anywhere
in function that is being normalized, so it applies to any subexpression as
well.

The \lam{expression conditions} list a number of conditions on the \lam{from}
expression that must hold for the transformation to apply.

Furthermore, there is some way to look into the environment (\eg, other top
level bindings).  The \lam{context conditions} part specifies any number of
top level bindings that must be present for the transformation to apply.
Usually, this lists a top level binding that binds an identfier that is also
used in the \lam{from} expression, allowing us to "access" the value of a top
level binding in the \lam{to} expression (\eg, for inlining).

Finally, there is a way to influence the environment. The \lam{context
additions} part lists any number of new top level bindings that should be
added.

If there are no \lam{context conditions} or \lam{context additions}, they can
be left out alltogether, along with the separator \lam{~}.

TODO: Example

\subsubsection{Other concepts}
A \emph{global variable} is any variable that is bound at the
top level of a program, or an external module. A local variable is any other
variable (\eg, variables local to a function, which can be bound by lambda
abstractions, let expressions and case expressions).

A \emph{hardware representable} type is a type that we can generate
a signal for in hardware. For example, a bit, a vector of bits, a 32 bit
unsigned word, etc. Types that are not runtime representable notably
include (but are not limited to): Types, dictionaries, functions.

A \emph{builtin function} is a function for which a builtin
hardware translation is available, because its actual definition is not
translatable. A user-defined function is any other function.

\subsubsection{Functions}
Here, we define a number of functions that can be used below to concisely
specify conditions.

\emph{gvar(expr)} is true when \emph{expr} is a variable that references a
global variable. It is false when it references a local variable.

\emph{lvar(expr)} is the inverse of \emph{gvar}; it is true when \emph{expr}
references a local variable, false when it references a global variable.

\emph{representable(expr)} or \emph{representable(var)} is true when
\emph{expr} or \emph{var} has a type that is representable at runtime.

\subsection{Normal form definition}
We can describe this normal form in a slightly more formal manner. The
following EBNF-like description completely captures the intended structure
(and generates a subset of GHC's core format).

Some clauses have an expression listed in parentheses. These are conditions
that need to apply to the clause.

\startlambda
\italic{normal} = \italic{lambda}
\italic{lambda} = λvar.\italic{lambda} (representable(var))
                | \italic{toplet} 
\italic{toplet} = let \italic{binding} in \italic{toplet} 
                | letrec [\italic{binding}] in \italic{toplet}
                | var (representable(varvar))
\italic{binding} = var = \italic{rhs} (representable(rhs))
                 -- State packing and unpacking by coercion
                 | var0 = var1 :: State ty (lvar(var1))
                 | var0 = var1 :: ty (var0 :: State ty) (lvar(var1))
\italic{rhs} = userapp
             | builtinapp
             -- Extractor case
             | case var of C a0 ... an -> ai (lvar(var))
             -- Selector case
             | case var of (lvar(var))
                DEFAULT -> var0 (lvar(var0))
                C w0 ... wn -> resvar (\forall{}i, wi \neq resvar, lvar(resvar))
\italic{userapp} = \italic{userfunc}
                 | \italic{userapp} {userarg}
\italic{userfunc} = var (gvar(var))
\italic{userarg} = var (lvar(var))
\italic{builtinapp} = \italic{builtinfunc}
                    | \italic{builtinapp} \italic{builtinarg}
\italic{builtinfunc} = var (bvar(var))
\italic{builtinarg} = \italic{coreexpr}
\stoplambda

-- TODO: Limit builtinarg further

-- TODO: There can still be other casts around (which the code can handle,
e.g., ignore), which still need to be documented here.

-- TODO: Note about the selector case. It just supports Bit and Bool
currently, perhaps it should be generalized in the normal form?

When looking at such a program from a hardware perspective, the top level
lambda's define the input ports. The value produced by the let expression is
the output port. Most function applications bound by the let expression
define a component instantiation, where the input and output ports are mapped
to local signals or arguments. Some of the others use a builtin
construction (\eg the \lam{case} statement) or call a builtin function
(\eg \lam{add} or \lam{sub}). For these, a hardcoded VHDL translation is
available.

\section{Transform passes}
In this section we describe the actual transforms. Here we're using
the core language in a notation that resembles lambda calculus.

Each of these transforms is meant to be applied to every (sub)expression
in a program, for as long as it applies. Only when none of the
expressions can be applied anymore, the program is in normal form. We
hope to be able to prove that this form will obey all of the constraints
defined above, but this has yet to happen (though it seems likely that
it will).

Each of the transforms will be described informally first, explaining
the need for and goal of the transform. Then, a formal definition is
given, using a familiar syntax from the world of logic. Each transform
is specified as a number of conditions (above the horizontal line) and a
number of conclusions (below the horizontal line). The details of using
this notation are still a bit fuzzy, so comments are welcom.

TODO: Formally describe the "apply to every (sub)expression" in terms of
rules with full transformations in the conditions.

\subsection{η-abstraction}
This transformation makes sure that all arguments of a function-typed
expression are named, by introducing lambda expressions. When combined with
β-reduction and function inlining below, all function-typed expressions should
be lambda abstractions or global identifiers.

\starttrans
E                 \lam{E :: * -> *}
--------------    \lam{E} is not the first argument of an application.
λx.E x            \lam{E} is not a lambda abstraction.
                  \lam{x} is a variable that does not occur free in \lam{E}.
\stoptrans

\startbuffer[from]
foo = λa.case a of 
  True -> λb.mul b b
  False -> id
\stopbuffer

\startbuffer[to]
foo = λa.λx.(case a of 
    True -> λb.mul b b
    False -> λy.id y) x
\stopbuffer

\transexample{η-abstraction}{from}{to}

\subsection{Extended β-reduction}
This transformation is meant to propagate application expressions downwards
into expressions as far as possible. In lambda calculus, this reduction
is known as β-reduction, but it is of course only defined for
applications of lambda abstractions. We extend this reduction to also
work for the rest of core (case and let expressions).

For let expressions:
\starttrans
let binds in E) M
-----------------
let binds in E M
\stoptrans

For case statements:
\starttrans
(case x of
  p1 -> E1
  \vdots
  pn -> En) M
-----------------
case x of
  p1 -> E1 M
  \vdots
  pn -> En M
\stoptrans

For lambda expressions:
\starttrans
(λx.E) M
-----------------
E[M/x]
\stoptrans

% And an example
\startbuffer[from]
( let a = (case x of 
            True -> id
            False -> neg
          ) 1
      b = (let y = 3 in add y) 2
  in
    (λz.add 1 z)
) 3
\stopbuffer

\startbuffer[to]
let a = case x of 
           True -> id 1
           False -> neg 1
    b = let y = 3 in add y 2
in
  add 1 3
\stopbuffer

\transexample{Extended β-reduction}{from}{to}

\subsection{Let derecursification}
This transformation is meant to make lets non-recursive whenever possible.
This might allow other optimizations to do their work better. TODO: Why is
this needed exactly?

\subsection{Let flattening}
This transformation puts nested lets in the same scope, by lifting the
binding(s) of the inner let into a new let around the outer let. Eventually,
this will cause all let bindings to appear in the same scope (they will all be
in scope for the function return value).

Note that this transformation does not try to be smart when faced with
recursive lets, it will just leave the lets recursive (possibly joining a
recursive and non-recursive let into a single recursive let). The let
rederursification transformation will do this instead.

\starttrans
letnonrec x = (let bindings in M) in N
------------------------------------------
let bindings in (letnonrec x = M) in N
\stoptrans

\starttrans
letrec 
  \vdots
  x = (let bindings in M)
  \vdots
in
  N
------------------------------------------
letrec
  \vdots
  bindings
  x = M
  \vdots
in
  N
\stoptrans

\startbuffer[from]
let
  a = letrec
    x = 1
    y = 2
  in
    x + y
in
  letrec
    b = let c = 3 in a + c
    d = 4
  in
    d + b
\stopbuffer
\startbuffer[to]
letrec
  x = 1
  y = 2
in
  let
    a = x + y
  in
    letrec
      c = 3
      b = a + c
      d = 4
    in
      d + b
\stopbuffer

\transexample{Let flattening}{from}{to}

\subsection{Empty let removal}
This transformation is simple: It removes recursive lets that have no bindings
(which usually occurs when let derecursification removes the last binding from
it).

\starttrans
letrec in M
--------------
M
\stoptrans

\subsection{Simple let binding removal}
This transformation inlines simple let bindings (\eg a = b).

This transformation is not needed to get into normal form, but makes the
resulting VHDL a lot shorter.

\starttrans
letnonrec
  a = b
in
  M
-----------------
M[b/a]
\stoptrans

\starttrans
letrec
  \vdots
  a = b
  \vdots
in
  M
-----------------
let
  \vdots [b/a]
  \vdots [b/a]
in
  M[b/a]
\stoptrans

\subsection{Unused let binding removal}
This transformation removes let bindings that are never used. Usually,
the desugarer introduces some unused let bindings.

This normalization pass should really be unneeded to get into normal form
(since ununsed bindings are not forbidden by the normal form), but in practice
the desugarer or simplifier emits some unused bindings that cannot be
normalized (e.g., calls to a \type{PatError} (TODO: Check this name)). Also,
this transformation makes the resulting VHDL a lot shorter.

\starttrans
let a = E in M
----------------------------    \lam{a} does not occur free in \lam{M}
M
\stoptrans

\starttrans
letrec
  \vdots
  a = E
  \vdots
in
  M
----------------------------    \lam{a} does not occur free in \lam{M}
letrec
  \vdots
  \vdots
in
  M
\stoptrans

\subsection{Non-representable binding inlining}
This transform inlines let bindings that have a non-representable type. Since
we can never generate a signal assignment for these bindings (we cannot
declare a signal assignment with a non-representable type, for obvious
reasons), we have no choice but to inline the binding to remove it.

If the binding is non-representable because it is a lambda abstraction, it is
likely that it will inlined into an application and β-reduction will remove
the lambda abstraction and turn it into a representable expression at the
inline site. The same holds for partial applications, which can be turned into
full applications by inlining.

Other cases of non-representable bindings we see in practice are primitive
Haskell types. In most cases, these will not result in a valid normalized
output, but then the input would have been invalid to start with. There is one
exception to this: When a builtin function is applied to a non-representable
expression, things might work out in some cases. For example, when you write a
literal \hs{SizedInt} in Haskell, like \hs{1 :: SizedInt D8}, this results in
the following core: \lam{fromInteger (smallInteger 10)}, where for example
\lam{10 :: GHC.Prim.Int\#} and \lam{smallInteger 10 :: Integer} have
non-representable types. TODO: This/these paragraph(s) should probably become a
separate discussion somewhere else.

\starttrans
letnonrec a = E in M
--------------------------    \lam{E} has a non-representable type.
M[E/a]
\stoptrans

\starttrans
letrec 
  \vdots
  a = E
  \vdots
in
  M
--------------------------    \lam{E} has a non-representable type.
letrec
  \vdots [E/a]
  \vdots [E/a]
in
  M[E/a]
\stoptrans

\startbuffer[from]
letrec
  a = smallInteger 10
  inc = λa -> add a 1
  inc' = add 1
  x = fromInteger a 
in
  inc (inc' x)
\stopbuffer

\startbuffer[to]
letrec
  x = fromInteger (smallInteger 10)
in
  (λa -> add a 1) (add 1 x)
\stopbuffer

\transexample{Let flattening}{from}{to}

\subsection{Compiler generated top level binding inlining}
TODO

\subsection{Scrutinee simplification}
This transform ensures that the scrutinee of a case expression is always
a simple variable reference.

\starttrans
case E of
  alts
-----------------        \lam{E} is not a local variable reference
let x = E in 
  case E of
    alts
\stoptrans

\startbuffer[from]
case (foo a) of
  True -> a
  False -> b
\stopbuffer

\startbuffer[to]
let x = foo a in
  case x of
    True -> a
    False -> b
\stopbuffer

\transexample{Let flattening}{from}{to}


\subsection{Case simplification}
This transformation ensures that all case expressions become normal form. This
means they will become one of:
\startitemize
\item An extractor case with a single alternative that picks a single field
from a datatype, \eg \lam{case x of (a, b) -> a}.
\item A selector case with multiple alternatives and only wild binders, that
makes a choice between expressions based on the constructor of another
expression, \eg \lam{case x of Low -> a; High -> b}.
\stopitemize

\starttrans
case E of
  C0 v0,0 ... v0,m -> E0
  \vdots
  Cn vn,0 ... vn,m -> En
--------------------------------------------------- \forall i \forall j, 0 <= i <= n, 0 <= i < m (\lam{wi,j} is a wild (unused) binder)
letnonrec
  v0,0 = case x of C0 v0,0 .. v0,m -> v0,0
  \vdots
  v0,m = case x of C0 v0,0 .. v0,m -> v0,m
  x0 = E0
  \dots
  vn,m = case x of Cn vn,0 .. vn,m -> vn,m
  xn = En
in
  case E of
    C0 w0,0 ... w0,m -> x0
    \vdots
    Cn wn,0 ... wn,m -> xn
\stoptrans

TODO: This transformation specified like this is complicated and misses
conditions to prevent looping with itself. Perhaps we should split it here for
discussion?

\startbuffer[from]
case a of
  True -> add b 1
  False -> add b 2
\stopbuffer

\startbuffer[to]
letnonrec
  x0 = add b 1
  x1 = add b 2
in
  case a of
    True -> x0
    False -> x1
\stopbuffer

\transexample{Selector case simplification}{from}{to}

\startbuffer[from]
case a of
  (,) b c -> add b c
\stopbuffer
\startbuffer[to]
letnonrec
  b = case a of (,) b c -> b
  c = case a of (,) b c -> c
  x0 = add b c
in
  case a of
    (,) w0 w1 -> x0
\stopbuffer

\transexample{Extractor case simplification}{from}{to}

\subsection{Case removal}
This transform removes any case statements with a single alternative and
only wild binders.

These "useless" case statements are usually leftovers from case simplification
on extractor case (see the previous example).

\starttrans
case x of
  C v0 ... vm -> E
----------------------     \lam{\forall i, 0 <= i <= m} (\lam{vi} does not occur free in E)
E
\stoptrans

\startbuffer[from]
case a of
  (,) w0 w1 -> x0
\stopbuffer

\startbuffer[to]
x0
\stopbuffer

\transexample{Case removal}{from}{to}

\subsection{Argument simplification}
The transforms in this section deal with simplifying application
arguments into normal form. The goal here is to:

\startitemize
 \item Make all arguments of user-defined functions (\eg, of which
 we have a function body) simple variable references of a runtime
 representable type. This is needed, since these applications will be turned
 into component instantiations.
 \item Make all arguments of builtin functions one of:
   \startitemize
    \item A type argument.
    \item A dictionary argument.
    \item A type level expression.
    \item A variable reference of a runtime representable type.
    \item A variable reference or partial application of a function type.
   \stopitemize
\stopitemize

When looking at the arguments of a user-defined function, we can
divide them into two categories:
\startitemize
  \item Arguments with a runtime representable type (\eg bits or vectors).

        These arguments can be preserved in the program, since they can
        be translated to input ports later on.  However, since we can
        only connect signals to input ports, these arguments must be
        reduced to simple variables (for which signals will be
        produced). This is taken care of by the argument extraction
        transform.
  \item Non-runtime representable typed arguments.
        
        These arguments cannot be preserved in the program, since we
        cannot represent them as input or output ports in the resulting
        VHDL. To remove them, we create a specialized version of the
        called function with these arguments filled in. This is done by
        the argument propagation transform.

        Typically, these arguments are type and dictionary arguments that are
        used to make functions polymorphic. By propagating these arguments, we
        are essentially doing the same which GHC does when it specializes
        functions: Creating multiple variants of the same function, one for
        each type for which it is used. Other common non-representable
        arguments are functions, e.g. when calling a higher order function
        with another function or a lambda abstraction as an argument.

        The reason for doing this is similar to the reasoning provided for
        the inlining of non-representable let bindings above. In fact, this
        argument propagation could be viewed as a form of cross-function
        inlining.
\stopitemize

TODO: Check the following itemization.

When looking at the arguments of a builtin function, we can divide them
into categories: 

\startitemize
  \item Arguments with a runtime representable type.
        
        As we have seen with user-defined functions, these arguments can
        always be reduced to a simple variable reference, by the
        argument extraction transform. Performing this transform for
        builtin functions as well, means that the translation of builtin
        functions can be limited to signal references, instead of
        needing to support all possible expressions.

  \item Arguments with a function type.
        
        These arguments are functions passed to higher order builtins,
        like \lam{map} and \lam{foldl}. Since implementing these
        functions for arbitrary function-typed expressions (\eg, lambda
        expressions) is rather comlex, we reduce these arguments to
        (partial applications of) global functions.
        
        We can still support arbitrary expressions from the user code,
        by creating a new global function containing that expression.
        This way, we can simply replace the argument with a reference to
        that new function. However, since the expression can contain any
        number of free variables we also have to include partial
        applications in our normal form.

        This category of arguments is handled by the function extraction
        transform.
  \item Other unrepresentable arguments.
        
        These arguments can take a few different forms:
        \startdesc{Type arguments}
          In the core language, type arguments can only take a single
          form: A type wrapped in the Type constructor. Also, there is
          nothing that can be done with type expressions, except for
          applying functions to them, so we can simply leave type
          arguments as they are.
        \stopdesc
        \startdesc{Dictionary arguments}
          In the core language, dictionary arguments are used to find
          operations operating on one of the type arguments (mostly for
          finding class methods). Since we will not actually evaluatie
          the function body for builtin functions and can generate
          code for builtin functions by just looking at the type
          arguments, these arguments can be ignored and left as they
          are.
        \stopdesc
        \startdesc{Type level arguments}
          Sometimes, we want to pass a value to a builtin function, but
          we need to know the value at compile time. Additionally, the
          value has an impact on the type of the function. This is
          encoded using type-level values, where the actual value of the
          argument is not important, but the type encodes some integer,
          for example. Since the value is not important, the actual form
          of the expression does not matter either and we can leave
          these arguments as they are.
        \stopdesc
        \startdesc{Other arguments}
          Technically, there is still a wide array of arguments that can
          be passed, but does not fall into any of the above categories.
          However, none of the supported builtin functions requires such
          an argument. This leaves use with passing unsupported types to
          a function, such as calling \lam{head} on a list of functions.

          In these cases, it would be impossible to generate hardware
          for such a function call anyway, so we can ignore these
          arguments.

          The only way to generate hardware for builtin functions with
          arguments like these, is to expand the function call into an
          equivalent core expression (\eg, expand map into a series of
          function applications). But for now, we choose to simply not
          support expressions like these.
        \stopdesc

        From the above, we can conclude that we can simply ignore these
        other unrepresentable arguments and focus on the first two
        categories instead.
\stopitemize

\subsubsection{Argument simplification}
This transform deals with arguments to functions that
are of a runtime representable type. It ensures that they will all become
references to global variables, or local signals in the resulting VHDL. 

TODO: It seems we can map an expression to a port, not only a signal.
Perhaps this makes this transformation not needed?
TODO: Say something about dataconstructors (without arguments, like True
or False), which are variable references of a runtime representable
type, but do not result in a signal.

To reduce a complex expression to a simple variable reference, we create
a new let expression around the application, which binds the complex
expression to a new variable. The original function is then applied to
this variable.

\starttrans
M N
--------------------    \lam{N} is of a representable type
let x = N in M x        \lam{N} is not a local variable reference
\stoptrans

\startbuffer[from]
add (add a 1) 1
\stopbuffer

\startbuffer[to]
let x = add a 1 in add x 1
\stopbuffer

\transexample{Argument extraction}{from}{to}

\subsubsection{Function extraction}
This transform deals with function-typed arguments to builtin functions.
Since these arguments cannot be propagated, we choose to extract them
into a new global function instead.

Any free variables occuring in the extracted arguments will become
parameters to the new global function. The original argument is replaced
with a reference to the new function, applied to any free variables from
the original argument.

This transformation is useful when applying higher order builtin functions
like \hs{map} to a lambda abstraction, for example. In this case, the code
that generates VHDL for \hs{map} only needs to handle top level functions and
partial applications, not any other expression (such as lambda abstractions or
even more complicated expressions).

\starttrans
M N                     \lam{M} is a (partial aplication of a) builtin function.
---------------------   \lam{f0 ... fn} = free local variables of \lam{N}
M x f0 ... fn           \lam{N :: a -> b}
~                       \lam{N} is not a (partial application of) a top level function
x = λf0 ... λfn.N
\stoptrans

\startbuffer[from]
map (λa . add a b) xs

map (add b) ys
\stopbuffer

\startbuffer[to]
x0 = λb.λa.add a b
~
map x0 xs

x1 = λb.add b
map x1 ys
\stopbuffer

\transexample{Function extraction}{from}{to}

\subsubsection{Argument propagation}
This transform deals with arguments to user-defined functions that are
not representable at runtime. This means these arguments cannot be
preserved in the final form and most be {\em propagated}.

Propagation means to create a specialized version of the called
function, with the propagated argument already filled in. As a simple
example, in the following program:

\startlambda
f = λa.λb.a + b
inc = λa.f a 1
\stoplambda

we could {\em propagate} the constant argument 1, with the following
result:

\startlambda
f' = λa.a + 1
inc = λa.f' a
\stoplambda

Special care must be taken when the to-be-propagated expression has any
free variables. If this is the case, the original argument should not be
removed alltogether, but replaced by all the free variables of the
expression. In this way, the original expression can still be evaluated
inside the new function. Also, this brings us closer to our goal: All
these free variables will be simple variable references.

To prevent us from propagating the same argument over and over, a simple
local variable reference is not propagated (since is has exactly one
free variable, itself, we would only replace that argument with itself).

This shows that any free local variables that are not runtime representable
cannot be brought into normal form by this transform. We rely on an
inlining transformation to replace such a variable with an expression we
can propagate again.

\starttrans
x = E
~
x Y0 ... Yi ... Yn                               \lam{Yi} is not of a runtime representable type
---------------------------------------------    \lam{Yi} is not a local variable reference
x' y0 ... yi-1 f0 ...  fm Yi+1 ... Yn            \lam{f0 ... fm} = free local vars of \lam{Yi}
~
x' = λy0 ... yi-1 f0 ... fm yi+1 ... yn .       
      E y0 ... yi-1 Yi yi+1 ... yn   

\stoptrans

TODO: Example

\subsection{Cast propagation / simplification}
This transform pushes casts down into the expression as far as possible. Since
its exact role and need is not clear yet, this transformation is not yet
specified.

\subsection{Return value simplification}
This transformation ensures that the return value of a function is always a
simple local variable reference.

Currently implemented using lambda simplification, let simplification, and
top simplification. Should change into something like the following, which
works only on the result of a function instead of any subexpression. This is
achieved by the contexts, like \lam{x = E}, though this is strictly not
correct (you could read this as "if there is any function \lam{x} that binds
\lam{E}, any \lam{E} can be transformed, while we only mean the \lam{E} that
is bound by \lam{x}. This might need some extra notes or something).

\starttrans
x = E                            \lam{E} is representable
~                                \lam{E} is not a lambda abstraction
E                                \lam{E} is not a let expression
---------------------------      \lam{E} is not a local variable reference
let x = E in x
\stoptrans

\starttrans
x = λv0 ... λvn.E
~                                \lam{E} is representable
E                                \lam{E} is not a let expression
---------------------------      \lam{E} is not a local variable reference
let x = E in x
\stoptrans

\starttrans
x = λv0 ... λvn.let ... in E
~                                \lam{E} is representable
E                                \lam{E} is not a local variable reference
---------------------------
let x = E in x
\stoptrans

\startbuffer[from]
x = add 1 2
\stopbuffer

\startbuffer[to]
x = let x = add 1 2 in x
\stopbuffer

\transexample{Return value simplification}{from}{to}