GitLab

Pointer equality is now defined using absolute object offsets. The treatment
is similar to CompCert:

* Equality of pointers in the same object is defined provided the object has
  not been deallocated.
* Equality of pointers in different objects is defined provided both pointers
  have not been deallocated and both are strict (i.e. not end-of-array).

Thus, pointer equality is defined for all pointers that are not-end-of-array
and have not been deallocated. The following examples have defined behavior:

  int x, y;
  printf("%d\n", &x == &y);
  int *p = malloc(sizeof(int)), *q = malloc(sizeof(int));
  printf("%d\n", p == q);
  struct S { int a; int b; } s, *r = &s;
  printf("%d\n", &s.a + 1 == &(r->b));

The following not:

  int x, y;
  printf("%d\n", &x + 1 == &y);

memory instead of the whole memory itself.

This has the following advantages:
* Avoid parametrization in {addresses,pointers,pointer_bits,bits}.v
* Make {base_values,values}.v independent of the memory, this makes better
  parallelized compilation possible.
* Allow small memories (e.g. singletons as used in separation logic) with
  addresses to objects in another part to be typed.
* Some proofs become easier, because the memory environments are preserved
  under many operations (insert, force, lock, unlock).

It also as the following disadvantages:
* At all kinds of places we now have explicit casts from memories to memory
  environments. This is kind of ugly. Note, we cannot declare memenv_of as a
  Coercion because it is non-uniform.
* It is a bit inefficient with respect to the interpreter, because memory
  environments are finite functions instead of proper functions, so calling
  memenv_of often (which we do) is not too good.

* Parametrize refinements with memories. This way, refinements imply typing,
  for example [w1 ⊑{Γ,f@m1↦m2} w2 : τ → (Γ,m1) ⊢ w1 : τ]. This relieves us from
  various hacks.
* Use addresses instead of index/references pairs for lookup and alter
  operations on memories.
* Prove various disjointness properties.

The refactoring includes:
* Use infix notations for the various list relations
* More consistent naming
* Put lemmas on one line whenever possible
* Change proofs into one-liners when possible
* Make better use of the "Implicit Types" command
* Improve the order of the list module by placing all definitions at the start,
  then the proofs, and finally the tactics.

Besides, there is some new machinery for proofs by reflection on lists. It is
used for a decision procedure for permutations and list containment.

Both the operational and axiomatic semantics are extended with sequence points
and a permission system based on fractional permissions. In order to achieve
this, the memory model has been completely revised, and is now built on top
of an abstract interface for permissions.

Apart from these changed, the library on lists and sets has been heavily
extended, and minor changed have been made to other parts of the prelude.

The development now corresponds exactly to the FoSSaCS 2013 paper.
Also, the prelude is updated to the one of the master branch.

* Define the standard strict order on pre orders.
* Prove that this strict order is well founded for finite sets and finite maps.
  We also provide some utilities to compute with well founded recursion.
* Improve the "simplify_option_equality" tactic to handle more cases.
* Axiomatize finiteness of finite maps by translation to lists, instead of by
  them having a finite domain.
* Prove many additional properties of finite maps.
* Add many functions and theorems on lists, including: permutations, resize,
  filter, ...

Most interestingly:
* Use [lia] instead of [omega] everywhere
* More many generic lemmas on the memory to the theory on finite maps.
* Many additional list lemmas.
* A new interface for a monad for collections, which is now also used by the
  collection tactics.
* Provide an additional finite collection implementation using unordered lists
  without duplicates removed. This implementation forms a monad (just the list
  monad in disguise).

The following things have been changed in this revision:

* We now give a small step semantics for expressions. The denotational semantics
  only works for side-effect free expressions.
* Dynamically allocated memory through alloc and free is now supported.
* The following expressions are added: assignment, function call, unary
  operators, conditional, alloc, and free.
* Some customary induction schemes for expressions are proven.
* The axiomatic semantics (and its interpretation) have been changed in order
  to deal with non-deterministic expressions.
* We have added inversion schemes based on small inversions for the operational
  semantics. Inversions using these schemes are much faster.
* We improved the statement preservation proof of the operational semantics.
* We now use a variant of SsReflect's [by] and [done], instead of Coq's [now]
  and [easy]. The [done] tactic is much faster as it does not perform
  inversions.
* Add theory, definitions and notations on vectors.
* Separate theory on contexts.
* Change [Arguments] declarations to ensure better unfolding.

improve some definitions, simplify some proofs.

The main changes are:

* Function calls in the operational semantics
* Mutually recursive function calls in the axiomatic semantics
* A general definition of the interpretation of the axiomatic semantics  so as
  to improve reusability (useful for function calls, and also for expressions
  in future versions)
* Type classes for stack independent, memory independent, and memory extensible
  assertions, and a lot of instances to automatically derive these properties.
* Many additional lemmas on the memory and more robust tactics to simplify
  goals involving is_free and mem_disjoint
* Proof of preservation of statements in the smallstep semantics

* Some new tactics: feed, feed destruct, feed inversion, etc...
* More robust tactic scripts using bullets and structured scripts
* Truncate most lines at 80 characters