Tree Intermediate Language (Tree-IL) is a structured intermediate language that is close in expressive power to Scheme. It is an expanded, pre-analyzed Scheme.
Tree-IL is “structured” in the sense that its representation is
based on records, not S-expressions. This gives a rigidity to the
language that ensures that compiling to a lower-level language only
requires a limited set of transformations. For example, the Tree-IL
<const> is a record type with two fields,
exp. Instances of this type are created via
Fields of this type are accessed via the
const-exp procedures. There is also a predicate,
See Records, for more information on records.
All Tree-IL types have a
src slot, which holds source location
information for the expression. This information, if present, will be
residualized into the compiled object code, allowing backtraces to
show source information. The format of
src is the same as that
returned by Guile’s
source-properties function. See Source Properties, for more information.
Although Tree-IL objects are represented internally using records,
there is also an equivalent S-expression external representation for
each kind of Tree-IL. For example, the S-expression representation
#<const src: #f exp: 3> expression would be:
Users may program with this format directly at the REPL:
scheme@(guile-user)> ,language tree-il Happy hacking with Tree Intermediate Language! To switch back, type `,L scheme'. tree-il@(guile-user)> (apply (primitive +) (const 32) (const 10)) ⇒ 42
src fields are left out of the external representation.
One may create Tree-IL objects from their external representations via
parse-tree-il, the reader for Tree-IL. If any source
information is attached to the input S-expression, it will be
propagated to the resulting Tree-IL expressions. This is probably the
easiest way to compile to Tree-IL: just make the appropriate external
representations in S-expression format, and let
take care of the rest.
An empty expression. In practice, equivalent to Scheme’s
A reference to a “primitive”. A primitive is a procedure that, when
compiled, may be open-coded. For example,
cons is usually
recognized as a primitive, so that it compiles down to a single
Compilation of Tree-IL usually begins with a pass that resolves some
<toplevel-ref> expressions to
<primitive-ref> expressions. The actual compilation pass
has special cases for applications of certain primitives, like
A reference to a lexically-bound variable. The name is the original name of the variable in the source program. gensym is a unique identifier for this variable.
Sets a lexically-bound variable.
A reference to a variable in a specific module. mod should be
the name of the module, e.g.
If public? is true, the variable named name will be looked
up in mod’s public interface, and serialized with
otherwise it will be looked up among the module’s private bindings,
and is serialized with
Sets a variable in a specific module.
References a variable from the current procedure’s module.
Sets a variable in the current procedure’s module.
Defines a new top-level variable in the current procedure’s module.
A conditional. Note that else is not optional.
A procedure call.
A closure. meta is an association list of properties for the
procedure. body is a single Tree-IL expression of type
<lambda-case>. As the
<lambda-case> clause can chain to
an alternate clause, this makes Tree-IL’s
<lambda> have the
expressiveness of Scheme’s
One clause of a
lambda expression in
Scheme is treated as a
case-lambda with one clause.
req is a list of the procedure’s required arguments, as symbols.
opt is a list of the optional arguments, or
#f if there
are no optional arguments. rest is the name of the rest
kw is a list of the form,
(keyword name var) ...), where keyword is the
keyword corresponding to the argument named name, and whose
corresponding gensym is var. inits are tree-il expressions
corresponding to all of the optional and keyword arguments, evaluated
to bind variables whose value is not supplied by the procedure caller.
Each init expression is evaluated in the lexical context of
previously bound variables, from left to right.
gensyms is a list of gensyms corresponding to all arguments: first all of the required arguments, then the optional arguments if any, then the rest argument if any, then all of the keyword arguments.
body is the body of the clause. If the procedure is called with
an appropriate number of arguments, body is evaluated in tail
position. Otherwise, if there is an alternate, it should be a
<lambda-case> expression, representing the next clause to try.
If there is no alternate, a wrong-number-of-arguments error is
Lexical binding, like Scheme’s
let. names are the
original binding names, gensyms are gensyms corresponding to the
names, and vals are Tree-IL expressions for the values.
exp is a single Tree-IL expression.
A version of
<let> that creates recursive bindings, like
letrec* if in-order? is true.
Dynamic binding; the equivalent of Scheme’s
fluids should be a list of Tree-IL expressions that will
evaluate to fluids, and vals a corresponding list of expressions
to bind to the fluids during the dynamic extent of the evaluation of
A dynamic variable reference. fluid should be a Tree-IL expression evaluating to a fluid.
A dynamic variable set. fluid, a Tree-IL expression evaluating to a fluid, will be set to the result of evaluating exp.
dynamic-wind. winder and unwinder should both
evaluate to thunks. Ensure that the winder and the unwinder are called
before entering and after leaving body. Note that body is
an expression, without a thunk wrapper.
A dynamic prompt. Instates a prompt named tag, an expression,
during the dynamic extent of the execution of body, also an
expression. If an abort occurs to this prompt, control will be passed
to handler, a
<lambda-case> expression with no optional
or keyword arguments, and no alternate. The first argument to the
<lambda-case> will be the captured continuation, and then all
of the values passed to the abort. See Prompts, for more
An abort to the nearest prompt with the name tag, an expression.
args should be a list of expressions to pass to the prompt’s
handler, and tail should be an expression that will evaluate to
a list of additional arguments. An abort will save the partial
continuation, which may later be reinstated, resulting in the
<abort> expression evaluating to some number of values.
There are two Tree-IL constructs that are not normally produced by higher-level compilers, but instead are generated during the source-to-source optimization and analysis passes that the Tree-IL compiler does. Users should not generate these expressions directly, unless they feel very clever, as the default analysis pass will generate them as necessary.
receive – binds the values returned by
exp to the
lambda-like bindings described by
gensyms. That is to say, gensyms may be an improper list.
<let-values> is an optimization of
<application> of the
<letrec>, but only for vals that are unset
fix is an optimization of
Tree-IL implements a compiler to GLIL that recursively traverses Tree-IL expressions, writing out GLIL expressions into a linear list. The compiler also keeps some state as to whether the current expression is in tail context, and whether its value will be used in future computations. This state allows the compiler not to emit code for constant expressions that will not be used (e.g. docstrings), and to perform tail calls when in tail position.
Most optimization, such as it currently is, is performed on Tree-IL expressions as source-to-source transformations. There will be more optimizations added in the future.
Interested readers are encouraged to read the implementation in
(language tree-il compile-glil) for more details.