Dafny

Dafny is an imperative and functional compiled language that compiles to other programming languages, such as C#, Java, JavaScript, Go and Python. It supports formal specification through preconditions, postconditions, loop invariants, loop variants, termination specifications and read/write framing specifications. The language combines ideas from the functional and imperative paradigms; it includes support for object-oriented programming. Features include generic classes, dynamic allocation, inductive datatypes and a variation of separation logic known as implicit dynamic frames for reasoning about side effects. Dafny was created by Rustan Leino at Microsoft Research after his previous work on developing ESC/Modula-3, ESC/Java, and Spec#.

Dafny is widely used in teaching because it provides a simple, integrated introduction to formal specification and verification; it is regularly featured in software verification competitions (e.g. VSTTE'08, VSCOMP'10, COST'11, and VerifyThis'12 ).

Dafny was designed as a verification-aware programming language, requiring verification along with code development. It thus fits the "Correct by Construction" software development paradigm. Verification proofs are supported by a mathematical toolbox that includes mathematical integers and reals, bit-vectors, sequences, sets, multisets, infinite sequences and sets, induction, co-induction, and calculational proofs. Verification obligations are discharged automatically, given sufficient specification. Dafny uses some program analysis to infer many specification assertions, reducing the burden on the user of writing specifications. The general proof framework is that of Hoare logic.

Dafny builds on the Boogie intermediate language which uses the Z3 automated theorem prover for discharging proof obligations.

Data types
Dafny provides methods for implementation which may have side-effects and functions for use in specification which are pure. Methods consist of sequences of statements following a familiar imperative style whilst, in contrast, the body of a function is simply an expression. Any side-effecting statements in a method (e.g. assigning an element of an array parameter) must be accounted for by noting which parameters can be mutated, using the  clause. Dafny also provides a range of immutable collection types including: sequences (e.g. ), sets (e.g.  ), maps, tuples, inductive datatypes and mutable arrays (e.g.  ).

Imperative features
The following illustrates many of the features in Dafny, including the use of preconditions, postconditions, loop invariants and loop variants.

This example computes the maximum element of an array. The method's precondition and postcondition are given with the  and   clauses (respectively). Likewise, the loop invariant and loop variant are given through the  and   clauses (respectively).

Loop invariants
The treatment of loop invariants in Dafny differs from traditional Hoare logic. Variables mutated in a loop are treated such that (most) information known about them prior to the loop is discarded. Information required to prove properties of such variables must be expressed explicitly in the loop invariant. In contrast, variables not mutated in the loop retain all information known about them beforehand. The following example illustrates using loops:

This fails verification because Dafny cannot establish that  holds at the assignment. From the precondition, intuitively,  holds in the loop since   is a NOP. However, this assignment causes Dafny to treat  as a mutable variable and drop information known about it from before the loop. To verify this program in Dafny we can either (a) remove the redundant assignment ; or (b) add the loop invariant

Dafny additionally employs limited static analysis to infer simple loop invariants where possible. In the example above, it would seem that the loop invariant  is also required as variable   is mutated within the loop. Whilst the underlying logic does require such an invariant, Dafny automatically infers this and, hence, it can be omitted at the source level.

Proof features
Dafny includes features which further support its use as a proof assistant. Although proofs of simple properties within a  or   (as shown above) are not unusual for tools of this nature, Dafny also allows the proof of properties between one   and another. As is common for a proof assistant, such proofs are often inductive in nature. Dafny is perhaps unusual in employing method invocation as a mechanism for applying the inductive hypothesis. The following illustrates:

Here,  proves a useful property between   and   (i.e. that  ). The use of a  for encoding lemmas and theorems is standard in Dafny with recursion employed for induction (typically, structural induction). Case analysis is performed using  statements and non-inductive cases are often discharged automatically. The verifier must also have complete access to the contents of a  or   in order to unroll them as necessary. This has implications when used in conjunction with access modifiers. Specifically, hiding the contents of a  using the   modifier can limit what properties can be established about it.