CHANGELOG.md 12.8 KB
Newer Older
Ralf Jung's avatar
Ralf Jung committed
1 2 3
In this changelog, we document "large-ish" changes to Iris that affect even the
way the logic is used on paper.  We also mention some significant changes in the
Coq development, but not every API-breaking change is listed.  Changes marked
4
`[#]` still need to be ported to the Iris Documentation LaTeX file(s).
5

6 7
## Iris master

Ralf Jung's avatar
Ralf Jung committed
8 9
Changes in and extensions of the theory:

Jacques-Henri Jourdan's avatar
Jacques-Henri Jourdan committed
10 11 12
* [#] Change in the definition of WP, so that there is a fancy update between
  the quantification over the next states and the later modality. This makes it
  possible to prove more powerful lifting lemmas: The new versions feature an
Ralf Jung's avatar
Ralf Jung committed
13
  "update that takes a step".
Ralf Jung's avatar
Ralf Jung committed
14 15 16
* [#] Weaken the semantics of CAS in heap_lang to be efficiently implementable:
  CAS may only be used to compare "unboxed" values that can be represented in a
  single machine word.
17 18 19
* [#] Add weakest preconditions for total program correctness.
* [#] "(Potentially) stuck" weakest preconditions are no longer considered
  experimental.
Ralf Jung's avatar
Ralf Jung committed
20

21 22
Changes in Coq:

Ralf Jung's avatar
Ralf Jung committed
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
* An all-new generalized proofmode that abstracts away from Iris!  The proofmode
  can now be used with non-step-indexed and even non-affine (i.e., linear)
  logics.  TODO: Add reference to paper.  Developments instantiating the
  proofmode typeclasses may need significant changes.  For developments just
  using the proofmode tactics, porting should not be too much effort.  Notable
  things to port are:
  - All the BI laws moved from the `uPred` module to the `bi` module.  For
    example, `uPred.later_equivI` became `bi.later_equivI`.
  - Big-ops are automatically imported, imports of `iris.base_logic.big_op` have
    to be removed.
  - The ⊢ notation can sometimes infer different (but convertible) terms when
    seraching for the BI to use, which (due to Coq limitations) can lead to
    failing rewrites, in particular when rewriting at function types.
* The `iInv` tactic can now be used without the second argument (the name for
  the closing update).  It will then instead add the obligation to close the
  invariant to the goal.
Ralf Jung's avatar
Ralf Jung committed
39 40
* Added support for defining derived connectives involving n-ary binders using
  telescopes.
Ralf Jung's avatar
Ralf Jung committed
41 42
* Improved pretty-printing of Iris connectives (in particular WP and fancy
  updates) when Coq has to line-wrap the output.
43
* Rename `timelessP` -> `timeless` (projection of the `Timeless` class)
Ralf Jung's avatar
Ralf Jung committed
44 45
* The CMRA axiom `cmra_extend` is now stated in `Type`, using `sigT` instead of
  in `Prop` using `exists`. This makes it possible to define the function space
Jacques-Henri Jourdan's avatar
Jacques-Henri Jourdan committed
46
  CMRA even for an infinite domain.
Robbert Krebbers's avatar
Robbert Krebbers committed
47 48 49 50 51
* Rename proof mode type classes for laters:
  - `IntoLaterN``MaybeIntoLaterN` (this one _may_ strip a later)
  - `IntoLaterN'``IntoLaterN` (this one _should_ strip a later)
  - `IntoLaterNEnv``MaybeIntoLaterNEnv`
  - `IntoLaterNEnvs``MaybeIntoLaterNEnvs`
52 53
* Rename:
  - `frag_auth_op``frac_auth_frag_op`
54 55 56
  - `cmra_opM_assoc``cmra_op_opM_assoc`
  - `cmra_opM_assoc_L``cmra_op_opM_assoc_L`
  - `cmra_opM_assoc'``cmra_opM_opM_assoc`
Robbert Krebbers's avatar
Robbert Krebbers committed
57
* `namespaces` has been moved to std++.
58 59
* Changed `IntoVal` to be directly usable for rewriting `e` into `of_val v`, and
  changed `AsVal` to be usable for rewriting via the `[v <-]` destruct pattern.
60

Ralf Jung's avatar
Ralf Jung committed
61
## Iris 3.1.0 (released 2017-12-19)
62

Ralf Jung's avatar
Ralf Jung committed
63 64
Changes in and extensions of the theory:

Ralf Jung's avatar
Ralf Jung committed
65
* Define `uPred` as a quotient on monotone predicates `M -> SProp`.
Ralf Jung's avatar
Ralf Jung committed
66 67
* Get rid of some primitive laws; they can be derived:
  `True ⊢ □ True` and `□ (P ∧ Q) ⊢ □ (P ∗ Q)`
68
* Camera morphisms have to be homomorphisms, not just monotone functions.
Ralf Jung's avatar
Ralf Jung committed
69
* Add a proof that `f` has a fixed point if `f^k` is contractive.
70 71
* Constructions for least and greatest fixed points over monotone predicates
  (defined in the logic of Iris using impredicative quantification).
Ralf Jung's avatar
Ralf Jung committed
72
* Add a proof of the inverse of `wp_bind`.
Ralf Jung's avatar
Ralf Jung committed
73
* [Experimental feature] Add new modality: ■ ("plainly").
Ralf Jung's avatar
Ralf Jung committed
74 75 76 77 78 79 80 81
* [Experimental feature] Support verifying code that might get stuck by
  distinguishing "non-stuck" vs. "(potentially) stuck" weakest
  preconditions. (See [Swasey et al., OOPSLA '17] for examples.) The non-stuck
  `WP e @ E {{ Φ }}` ensures that, as `e` runs, it does not get stuck. The stuck
  `WP e @ E ?{{ Φ }}` ensures that, as usual, all invariants are preserved while
  `e` runs, but it permits execution to get stuck. The former implies the
  latter. The full judgment is `WP e @ s; E {{ Φ }}`, where non-stuck WP uses
  *stuckness bit* `s = NotStuck` while stuck WP uses `s = MaybeStuck`.
Ralf Jung's avatar
Ralf Jung committed
82 83 84

Changes in Coq:

Robbert's avatar
Robbert committed
85 86
* Move the `prelude` folder to its own project:
  [coq-std++](https://gitlab.mpi-sws.org/robbertkrebbers/coq-stdpp)
Ralf Jung's avatar
Ralf Jung committed
87 88 89 90 91 92 93 94 95 96 97
* Some extensions/improvements of heap_lang:
  - Improve handling of pure (non-state-dependent) reductions.
  - Add fetch-and-add (`FAA`) operation.
  - Add syntax for all Coq's binary operations on `Z`.
* Generalize `saved_prop` to let the user choose the location of the type-level
  later.  Rename the general form to `saved_anything`.  Provide `saved_prop` and
  `saved_pred` as special cases.
* Improved big operators:
  + They are no longer tied to cameras, but work on any monoid
  + The version of big operations over lists was redefined so that it enjoys
    more definitional equalities.
Ralf Jung's avatar
Ralf Jung committed
98
* Rename some things and change notation:
99
  - The unit of a camera: `empty` -> `unit`, `∅` -> `ε`
100
  - Disjointness: `⊥` -> `##`
Ralf Jung's avatar
Ralf Jung committed
101
  - A proof mode type class `IntoOp` -> `IsOp`
102 103 104 105 106 107
  - OFEs with all elements being discrete: `Discrete` -> `OfeDiscrete`
  - OFE elements whose equality is discrete: `Timeless` -> `Discrete`
  - Timeless propositions: `TimelessP` -> `Timeless`
  - Camera elements such that `core x = x`: `Persistent` -> `CoreId`
  - Persistent propositions: `PersistentP` -> `Persistent`
  - The persistent modality: `always` -> `persistently`
108
  - Adequacy for non-stuck weakestpre: `adequate_safe` -> `adequate_not_stuck`
Ralf Jung's avatar
Ralf Jung committed
109
  - Consistently SnakeCase identifiers:
110 111 112 113 114 115 116 117 118 119
    + `CMRAMixin` -> `CmraMixin`
    + `CMRAT` -> `CmraT`
    + `CMRATotal` -> `CmraTotal`
    + `CMRAMorphism` -> `CmraMorphism`
    + `CMRADiscrete` -> `CmraDiscrete`
    + `UCMRAMixin` -> `UcmraMixin`
    + `UCMRAT` -> `UcmraT`
    + `DRAMixin` -> `DraMixin`
    + `DRAT` -> `DraT`
    + `STS` -> `Sts`
Ralf Jung's avatar
Ralf Jung committed
120 121
  - Many lemmas also changed their name.  `always_*` became `persistently_*`,
    and furthermore: (the following list is not complete)
122 123 124 125 126 127 128 129 130 131 132
    + `impl_wand` -> `impl_wand_1` (it only involves one direction of the
      equivalent)
    + `always_impl_wand` -> `impl_wand`
    + `always_and_sep_l` -> `and_sep_l`
    + `always_and_sep_r` -> `and_sep_r`
    + `always_sep_dup` -> `sep_dup`
    + `wand_impl_always` -> `impl_wand_persistently` (additionally,
      the direction of this equivalence got swapped for consistency's sake)
    + `always_wand_impl` -> `persistently_impl_wand` (additionally, the
      direction of this equivalence got swapped for consistency's sake)
  The following `sed` snippet should get you most of the way:
Ralf Jung's avatar
Ralf Jung committed
133
```
Ralf Jung's avatar
Ralf Jung committed
134
sed 's/\bPersistentP\b/Persistent/g; s/\bTimelessP\b/Timeless/g; s/\bCMRADiscrete\b/CmraDiscrete/g; s/\bCMRAT\b/CmraT/g; s/\bCMRAMixin\b/CmraMixin/g; s/\bUCMRAT\b/UcmraT/g; s/\bUCMRAMixin\b/UcmraMixin/g; s/\bSTS\b/Sts/g' -i $(find -name "*.v")
Ralf Jung's avatar
Ralf Jung committed
135
```
Ralf Jung's avatar
Ralf Jung committed
136 137
* `PersistentL` and `TimelessL` (persistence and timelessness of lists of
  propositions) are replaces by `TCForall` from std++.
Ralf Jung's avatar
Ralf Jung committed
138 139
* Fix a bunch of consistency issues in the proof mode, and make it overall more
  usable.  In particular:
140 141 142 143 144 145 146 147 148 149
  - All proof mode tactics start the proof mode if necessary; `iStartProof` is
    no longer needed and should only be used for building custom proof mode
    tactics.
  - Change in the grammar of specialization patterns: `>[...]` -> `[> ...]`
  - Various new specification patterns for `done` and framing.
  - There is common machinery for symbolic execution of pure reductions. This
    is provided by the type classes `PureExec` and `IntoVal`.
  - There is a new connective `tc_opaque`, which can be used to make definitions
    opaque for type classes, and thus opaque for most tactics of the proof
    mode.
Ralf Jung's avatar
Ralf Jung committed
150 151 152 153 154 155 156 157
  - Define Many missing type class instances for distributing connectives.
  - Implement the tactics `iIntros (?)` and `iIntros "!#"` (i.e. `iAlways`)
    using type classes. This makes them more generic, e.g., `iIntros (?)` also
    works when the universal quantifier is below a modality, and `iAlways` also
    works for the plainness modality.  A breaking change, however, is that these
    tactics now no longer work when the universal quantifier or modality is
    behind a type class opaque definition.  Furthermore, this can change the
    name of anonymous identifiers introduced with the "%" pattern.
Ralf Jung's avatar
Ralf Jung committed
158
* Make `ofe_fun` dependently typed, subsuming `iprod`.  The latter got removed.
Ralf Jung's avatar
Ralf Jung committed
159 160 161
* Define the generic `fill` operation of the `ectxi_language` construct in terms
  of a left fold instead of a right fold. This gives rise to more definitional
  equalities.
Robbert Krebbers's avatar
Robbert Krebbers committed
162 163 164 165 166
* The language hierarchy (`language`, `ectx_language`, `ectxi_language`) is now
  fully formalized using canonical structures instead of using a mixture of
  type classes and canonical structures. Also, it now uses explicit mixins. The
  file `program_logic/ectxi_language` contains some documentation on how to
  setup Iris for your language.
Ralf Jung's avatar
Ralf Jung committed
167 168
* Restore the original, stronger notion of atomicity alongside the weaker
  notion. These are `Atomic a e` where the stuckness bit `s` indicates whether
169 170
  expression `e` is weakly (`a = WeaklyAtomic`) or strongly
  (`a = StronglyAtomic`) atomic.
171 172 173 174
* Various improvements to `solve_ndisj`.
* Use `Hint Mode` to prevent Coq from making arbitrary guesses in the presence
  of evars, which often led to divergence. There are a few places where type
  annotations are now needed.
175
* The rules `internal_eq_rewrite` and `internal_eq_rewrite_contractive` are now
Ralf Jung's avatar
Ralf Jung committed
176
  stated in the logic, i.e., they are `iApply`-friendly.
177

Ralf Jung's avatar
Ralf Jung committed
178
## Iris 3.0.0 (released 2017-01-11)
179

Ralf Jung's avatar
Ralf Jung committed
180 181 182
* There now is a deprecation process.  The modules `*.deprecated` contain
  deprecated notations and definitions that are provided for backwards
  compatibility and will be removed in a future version of Iris.
Ralf Jung's avatar
Ralf Jung committed
183
* View shifts are radically simplified to just internalize frame-preserving
184
  updates.  Weakestpre is defined inside the logic, and invariants and view
Ralf Jung's avatar
Ralf Jung committed
185
  shifts with masks are also coded up inside Iris.  Adequacy of weakestpre is
186
  proven in the logic. The old ownership of the entire physical state is
Ralf Jung's avatar
Ralf Jung committed
187 188
  replaced by a user-selected predicate over physical state that is maintained
  by weakestpre.
189
* Use OFEs instead of COFEs everywhere.  COFEs are only used for solving the
190 191
  recursive domain equation.  As a consequence, CMRAs no longer need a proof of
  completeness.  (The old `cofeT` is provided by `algebra.deprecated`.)
192
* Implement a new agreement construction.  Unlike the old one, this one
193 194
  preserves discreteness.  dec_agree is thus no longer needed and has been moved
  to algebra.deprecated.
195 196 197
* Renaming and moving things around: uPred and the rest of the base logic are in
  `base_logic`, while `program_logic` is for everything involving the general
  Iris notion of a language.
198 199 200 201
* Renaming in prelude.list: Rename `prefix_of` -> `prefix` and `suffix_of` ->
  `suffix` in lemma names, but keep notation ``l1 `prefix_of` l2`` and ``l1
  `suffix_of` l2``.  `` l1 `sublist` l2`` becomes ``l1 `sublist_of` l2``. Rename
  `contains` -> `submseteq` and change `` l1 `contains` l2`` to ``l1 ⊆+ l2``.
Ralf Jung's avatar
Ralf Jung committed
202 203 204
* Slightly weaker notion of atomicity: an expression is atomic if it reduces in
  one step to something that does not reduce further.
* Changed notation for embedding Coq assertions into Iris.  The new notation is
205
  ⌜φ⌝.  Also removed `=` and `⊥` from the Iris scope.  (The old notations are
Ralf Jung's avatar
Ralf Jung committed
206
  provided in `base_logic.deprecated`.)
207
* Up-closure of namespaces is now a notation (↑) instead of a coercion.
208 209
* With invariants and the physical state being handled in the logic, there is no
  longer any reason to demand the CMRA unit to be discrete.
Ralf Jung's avatar
Ralf Jung committed
210
* The language can now fork off multiple threads at once.
211 212
* Local Updates (for the authoritative monoid) are now a 4-way relation with
  syntax-directed lemmas proving them.
Ralf Jung's avatar
Ralf Jung committed
213 214 215

## Iris 2.0

Ralf Jung's avatar
Ralf Jung committed
216 217 218 219
* [heap_lang] No longer use dependent types for expressions.  Instead, values
  carry a proof of closedness.  Substitution, closedness and value-ness proofs
  are performed by computation after reflecting into a term langauge that knows
  about values and closed expressions.
220 221
* [program_logic/language] The language does not define its own "atomic"
  predicate.  Instead, atomicity is defined as reducing in one step to a value.
222 223 224 225 226
* [program_logic] Due to a lack of maintenance and usefulness, lifting lemmas
  for Hoare triples are removed.

## Iris 2.0-rc2

Ralf Jung's avatar
Ralf Jung committed
227
This version matches the final ICFP 2016 paper.
228 229

* [algebra] Make the core of an RA or CMRA a partial function.
Ralf Jung's avatar
Ralf Jung committed
230
* [program_logic/lifting] Lifting lemmas no longer round-trip through a
Ralf Jung's avatar
Ralf Jung committed
231 232 233
  user-chosen predicate to define the configurations we can reduce to; they
  directly relate to the operational semantics.  This is equivalent and
  much simpler to read.
Ralf Jung's avatar
Ralf Jung committed
234 235 236

## Iris 2.0-rc1

Ralf Jung's avatar
Ralf Jung committed
237
This is the Coq development and Iris Documentation as submitted to ICFP 2016.