Our next major topic is type systems -- static program
analyses that classify expressions according to the shapes
of
their results. We'll begin with a typed version of the simplest
imaginable language, to introduce the basic ideas of types and
typing rules and the fundamental theorems about type systems:
type preservation and progress. In chapter Stlc we'll move
on to the simply typed lambda-calculus, which lives at the core
of every modern functional programming language (including
Coq!).
To motivate the discussion of type systems, let's begin as usual with a tiny toy language. We want it to have the potential for programs to go wrong because of runtime type errors, so we need something a tiny bit more complex than the language of constants and addition that we used in chapter Smallstep: a single kind of data (e.g., numbers) is too simple, but just two kinds (numbers and booleans) gives us enough material to tell an interesting story.
The language definition is completely routine.
Here is the syntax, informally:
t ::= tru | fls | test t then t else t | zro | scc t | prd t | iszro t
And here it is formally:
Values are tru, fls, and numeric values...
Here is the single-step relation, first informally...
t1 --> t1'
t1 --> t1'
numeric value v1
t1 --> t1'
numeric value v1
t1 --> t1'
... and then formally:
Notice that the step relation doesn't care about whether the expression being stepped makes global sense -- it just checks that the operation in the next reduction step is being applied to the right kinds of operands. For example, the term scc tru cannot take a step, but the almost as obviously nonsensical term
scc (test tru then tru else tru)
can take a step (once, before becoming stuck).
The first interesting thing to notice about this step relation
is that the strong progress theorem from the Smallstep chapter
fails here. That is, there are terms that are normal forms (they
can't take a step) but not values (because we have not included
them in our definition of possible results of reduction
). Such
terms are stuck.
However, although values and normal forms are not the same in this language, the set of values is a subset of the set of normal forms. This is important because it shows we did not accidentally define things so that some value could still take a step.
(Hint: You will reach a point in this proof where you need to use an induction to reason about a term that is known to be a numeric value. This induction can be performed either over the term itself or over the evidence that it is a numeric value. The proof goes through in either case, but you will find that one way is quite a bit shorter than the other. For the sake of the exercise, try to complete the proof both ways.)
Use value_is_nf to show that the step relation is also deterministic.
The next critical observation is that, although this language has stuck terms, they are always nonsensical, mixing booleans and numbers in a way that we don't even want to have a meaning. We can easily exclude such ill-typed terms by defining a typing relation that relates terms to the types (either numeric or boolean) of their final results.
In informal notation, the typing relation is often written
|- t \in T and pronounced t has type T.
The |- symbol
is called a turnstile.
Below, we're going to see richer typing
relations where one or more additional context
arguments are
written to the left of the turnstile. For the moment, the context
is always empty.
|- t1 \in Bool |- t2 \in T |- t3 \in T
|- t1 \in Nat
|- t1 \in Nat
|- t1 \in Nat
(Since we've included all the constructors of the typing relation in the hint database, the auto tactic can actually find this proof automatically.)
It's important to realize that the typing relation is a conservative (or static) approximation: it does not consider what happens when the term is reduced -- in particular, it does not calculate the type of its normal form.
The following two lemmas capture the fundamental property that the definitions of boolean and numeric values agree with the typing relation.
The typing relation enjoys two critical properties. The first is that well-typed normal forms are not stuck -- or conversely, if a term is well typed, then either it is a value or it can take at least one step. We call this progress.
Complete the formal proof of the progress property. (Make sure you understand the parts we've given of the informal proof in the following exercise before starting -- this will save you a lot of time.)
Complete the corresponding informal proof:
Theorem: If |- t \in T, then either t is a value or else t --> t' for some t'.
Proof: By induction on a derivation of |- t \in T.
This theorem is more interesting than the strong progress theorem that we saw in the Smallstep chapter, where all normal forms were values. Here a term can be stuck, but only if it is ill typed.
The second critical property of typing is that, when a well-typed term takes a step, the result is also a well-typed term.
Complete the formal proof of the preservation property. (Again, make sure you understand the informal proof fragment in the following exercise first.)
Complete the following informal proof:
Theorem: If |- t \in T and t --> t', then |- t' \in T.
Proof: By induction on a derivation of |- t \in T.
Inspecting the rules for the small-step reduction relation and remembering that t has the form test ..., we see that the only ones that could have been used to prove t --> t' are ST_TestTru, ST_TestFls, or ST_Test.
Now prove the same property again by induction on the evaluation derivation instead of on the typing derivation. Begin by carefully reading and thinking about the first few lines of the above proofs to make sure you understand what each one is doing. The set-up for this proof is similar, but not exactly the same.
The preservation theorem is often called subject reduction,
because it tells us what happens when the subject
of the typing
relation is reduced. This terminology comes from thinking of
typing statements as sentences, where the term is the subject and
the type is the predicate.
Putting progress and preservation together, we see that a well-typed term can never reach a stuck state.
Having seen the subject reduction property, one might wonder whether the opposity property -- subject expansion -- also holds. That is, is it always the case that, if t --> t' and |- t' \in T, then |- t \in T? If so, prove it. If not, give a counter-example. (You do not need to prove your counter-example in Coq, but feel free to do so.)
(* FILL IN HERE *)
Suppose that we add this new rule to the typing relation:
| T_SccBool : forall t, |- t \in Bool -> |- scc t \in Bool
Which of the following properties remain true in the presence of
this rule? For each one, write either remains true
or
else becomes false.
If a property becomes false, give a
counterexample.
Suppose, instead, that we add this new rule to the step relation:
| ST_Funny1 : forall t2 t3, (test tru t2 t3) --> t3
Which of the above properties become false in the presence of this rule? For each one that does, give a counter-example. (* FILL IN HERE *)
Suppose instead that we add this rule:
| ST_Funny2 : forall t1 t2 t2' t3, t2 --> t2' -> (test t1 t2 t3) --> (test t1 t2' t3)
Which of the above properties become false in the presence of this rule? For each one that does, give a counter-example. (* FILL IN HERE *)
Suppose instead that we add this rule:
| ST_Funny3 : (prd fls) --> (prd (prd fls))
Which of the above properties become false in the presence of this rule? For each one that does, give a counter-example. (* FILL IN HERE *)
Suppose instead that we add this rule:
| T_Funny4 : |- zro \in Bool
Which of the above properties become false in the presence of this rule? For each one that does, give a counter-example. (* FILL IN HERE *)
Suppose instead that we add this rule:
| T_Funny5 : |- prd zro \in Bool
Which of the above properties become false in the presence of this rule? For each one that does, give a counter-example. (* FILL IN HERE *)
Make up some exercises of your own along the same lines as the ones above. Try to find ways of selectively breaking properties -- i.e., ways of changing the definitions that break just one of the properties and leave the others alone.
The reduction rule ST_PrdZro is a bit counter-intuitive: we might feel that it makes more sense for the predecessor of zro to be undefined, rather than being defined to be zro. Can we achieve this simply by removing the rule from the definition of step? Would doing so create any problems elsewhere?
(* FILL IN HERE *)
Suppose our evaluation relation is defined in the big-step style. State appropriate analogs of the progress and preservation properties. (You do not need to prove them.)
Can you see any limitations of either of your properties? Do they allow for nonterminating commands? Why might we prefer the small-step semantics for stating preservation and progress?
(* FILL IN HERE *)