diff options
| author | Gaëtan Gilbert | 2020-04-28 20:29:33 +0200 |
|---|---|---|
| committer | Gaëtan Gilbert | 2020-04-28 20:29:33 +0200 |
| commit | 0e46754c30573206299c0ef2cbf2289a592bbcda (patch) | |
| tree | 0349be278f15aaad42fa58b73d509e0d5bee9710 | |
| parent | 16559843925f3489b61920ff398680f10f1f00cc (diff) | |
| parent | 0b16247d6b9bb54c8c4ead42ecacfb0f59396197 (diff) | |
Merge PR #12183: Suggestion of improvement for the Allow SProp error message.
Reviewed-by: SkySkimmer
Reviewed-by: jfehrle
| -rw-r--r-- | doc/sphinx/addendum/sprop.rst | 45 | ||||
| -rw-r--r-- | vernac/himsg.ml | 6 |
2 files changed, 28 insertions, 23 deletions
diff --git a/doc/sphinx/addendum/sprop.rst b/doc/sphinx/addendum/sprop.rst index b2d3687780..b19239ed22 100644 --- a/doc/sphinx/addendum/sprop.rst +++ b/doc/sphinx/addendum/sprop.rst @@ -7,27 +7,26 @@ SProp (proof irrelevant propositions) The status of strict propositions is experimental. + In particular, conversion checking through bytecode or native code + compilation currently does not understand proof irrelevance. + This section describes the extension of |Coq| with definitionally proof irrelevant propositions (types in the sort :math:`\SProp`, also known as strict propositions) as described in :cite:`Gilbert:POPL2019`. -Using :math:`\SProp` may be prevented by passing ``-disallow-sprop`` -to the |Coq| program or using :flag:`Allow StrictProp`. +Use of |SProp| may be disabled by passing ``-disallow-sprop`` to the +|Coq| program or by turning the :flag:`Allow StrictProp` flag off. .. flag:: Allow StrictProp :name: Allow StrictProp - Allows using :math:`\SProp` when set and forbids it when unset. The - initial value depends on whether you used the command line - ``-disallow-sprop`` and ``-allow-sprop``. - -.. exn:: SProp not allowed, you need to Set Allow StrictProp or to use the -allow-sprop command-line-flag. - :undocumented: - -.. coqtop:: none + Enables or disables the use of |SProp|. It is enabled by default. + The command-line flag ``-disallow-sprop`` disables |SProp| at + startup. - Set Allow StrictProp. + .. exn:: SProp is disallowed because the "Allow StrictProp" flag is off. + :undocumented: Some of the definitions described in this document are available through ``Coq.Logic.StrictProp``, which see. @@ -38,29 +37,35 @@ Basic constructs The purpose of :math:`\SProp` is to provide types where all elements are convertible: -.. coqdoc:: +.. coqtop:: all - Definition irrelevance (A:SProp) (P:A -> Prop) (x:A) (v:P x) (y:A) : P y := v. + Theorem irrelevance (A : SProp) (P : A -> Prop) : forall x : A, P x -> forall y : A, P y. + Proof. + intros * Hx *. + exact Hx. + Qed. Since we have definitional :ref:`eta-expansion` for functions, the property of being a type of definitionally irrelevant values is impredicative, and so is :math:`\SProp`: -.. coqdoc:: +.. coqtop:: all Check fun (A:Type) (B:A -> SProp) => (forall x:A, B x) : SProp. -.. warning:: - - Conversion checking through bytecode or native code compilation - currently does not understand proof irrelevance. - In order to keep conversion tractable, cumulativity for :math:`\SProp` -is forbidden: +is forbidden, unless the :flag:`Cumulative StrictProp` flag is turned +on: .. coqtop:: all Fail Check (fun (A:SProp) => A : Type). + Set Cumulative StrictProp. + Check (fun (A:SProp) => A : Type). + +.. coqtop:: none + + Unset Cumulative StrictProp. We can explicitly lift strict propositions into the relevant world by using a wrapping inductive type. The inductive stops definitional diff --git a/vernac/himsg.ml b/vernac/himsg.ml index fddc84b398..41f2ab9c63 100644 --- a/vernac/himsg.ml +++ b/vernac/himsg.ml @@ -729,9 +729,9 @@ let explain_undeclared_universe env sigma l = spc () ++ str "(maybe a bugged tactic)." let explain_disallowed_sprop () = - Pp.(strbrk "SProp not allowed, you need to " - ++ str "Set Allow StrictProp" - ++ strbrk " or to use the -allow-sprop command-line-flag.") + Pp.(strbrk "SProp is disallowed because the " + ++ str "\"Allow StrictProp\"" + ++ strbrk " flag is off.") let explain_bad_relevance env = strbrk "Bad relevance (maybe a bugged tactic)." |
