GitLab

Now we try to avoid adding them unnecessarily, so we don't have to remove them automatically any more.

And also rename the corresponding proof mode tactics.

This fact is deduced from reducibility. Unfortunately, this sometimes depends on the type of states being inhabited, so that this additional hypothesis sometimes appear.

rename program_logic.{ownership -> wsat}. It really is about world satisfaction and invariants more than about ownership.

This also removes the double use of the name 'wp_fork' in both
program_logic/weakestpre and heap_lang/lifting.

Also, since do_head_step no longer has a purpose, I have removed it
and just use a bunch of eauto hints.

This is more consistent with CAS, which also can be used on any value.
Note that being able to (atomically) test for equality of any value and
being able to CAS on any value is not realistic. See the discussion at
https://gitlab.mpi-sws.org/FP/iris-coq/issues/26, and in particular JH
Jourdan's observation:

  I think indeed for heap_lang this is just too complicated.

  Anyway, the role of heap_lang is not to model any actual
  programming language, but rather to show that we can do proofs
  about certain programs. The fact that you can write unrealistic
  programs is not a problem, IMHO. The only thing which is important
  is that the program that we write are realistic (i.e., faithfully
  represents the algorithm we want to p

This commit is based on a commit by Zhen Zhang who generalized equality
to work on any literal (and not just integers).

This generalization is surprisingly easy in Iris 3.0, so I could not
resist not doing it :).

This better reflects the name of the bind rule.

I renamed an internal tactic that was previously called wp_bind into
wp_bind_core.

This commit features:

- A simpler model. The recursive domain equation no longer involves a triple
  containing invariants, physical state and ghost state, but just ghost state.
  Invariants and physical state are encoded using (higher-order) ghost state.

- (Primitive) view shifts are formalized in the logic and all properties about
  it are proven in the logic instead of the model. Instead, the core logic
  features only a notion of raw view shifts which internalizing performing frame
  preserving updates.

- A better behaved notion of mask changing view shifts. In particular, we no
  longer have side-conditions on transitivity of view shifts, and we have a
  rule for introduction of mask changing view shifts |={E1,E2}=> P with
  E2 ⊆ E1 which allows to postpone performing a view shift.

- The weakest precondition connective is formalized in the logic using Banach's
  fixpoint. All properties about the connective are proven in the logic instead
  of directly in the model.

- Adequacy is proven in the logic and uses a primitive form of adequacy for
  uPred that only involves raw views shifts and laters.

Some remarks:

- I have removed binary view shifts. I did not see a way to describe all rules
  of the new mask changing view shifts using those.
- There is no longer the need for the notion of "frame shifting assertions" and
  these are thus removed. The rules for Hoare triples are thus also stated in
  terms of primitive view shifts.

TODO:

- Maybe rename primitive view shift into something more sensible
- Figure out a way to deal with closed proofs (see the commented out stuff in
  tests/heap_lang and tests/barrier_client).

This reverts commit 20b4ae55, which does
not seem to work with Coq 8.5pl2 (I accidentally tested with 8.5pl1).

This makes type checking more directed, and somewhat more predictable.

On the downside, it makes it impossible to declare the singleton on
lists as an instance of SingletonM and the insert and alter operations
on functions as instances of Alter and Insert. However, these were not
used often anyway.

* Values are considered as atomic expressions (this does not hurt, and
  this makes the proofs of atomicity simpler).

The intropattern {H} also meant clear (both in ssreflect, and the logic
part of the introduction pattern).

I know we don't use it.  Stating theorems also serves to document things, and IMHO this one is informative.  It also costs us nothing.

Concretely, when execution of any of the wp_ tactics does not yield
another wp, it will make sure that a view shift is kept. This
behavior was already partially there, but now it is hopefully more
consistent.

Changes:
- We no longer have a different syntax for specializing a term H : P -★ Q whose
  range P or domain Q is persistent. There is just one syntax, and the system
  automatically determines whether either P or Q is persistent.
- While specializing a term, always modalities are automatically stripped. This
  gets rid of the specialization pattern !.
- Make the syntax of specialization patterns more consistent. The syntax for
  generating a goal is [goal_spec] where goal_spec is one of the following:

    H1 .. Hn : generate a goal using hypotheses H1 .. Hn
   -H1 .. Hn : generate a goal using all hypotheses but H1 .. Hn
           # : generate a goal for the premise in which all hypotheses can be
               used. This is only allowed when specializing H : P -★ Q where
               either P or Q is persistent.
           % : generate a goal for a pure premise.

This gets rid of the (ambiguous) notation %l, because we can declare
LitLoc as a coercion. It also shortens the code.

That way, we do not have useless type annotations of the form
"v : language.val heap_lang" cluttering about any goal.

Note, that we could decide to eta expand everywhere (as we do for ∀
and ∃), and use the notation "WP e {{ Q }}" for "wp e ⊤ (λ _, Q)".

and show that this is an instance of evaluation contexts

This required a new ectx axiom: Positivity of evaluation contexts. This axiom was
also present in the old Iris 1.1 development, back when it still derived lifting
axioms for ectx languages.