Add elaboration of qualities by TDiazT · Pull Request #21450 · rocq-prover/rocq

TDiazT · 2025-12-22T19:07:54Z

This PR depends on #21417 .

This PR adds elaboration of (implicit) qualities, which is enabled by setting the flag Universe Polymorphism and disabling a new flag Collapse Sorts ToType.

Associated RFC : rocq-prover/rfcs#111.

Examples

The test file test-suite/success/sort_poly_elab.v has been updated with more examples.

Set Universe Polymorphism.
Unset Collapse Sorts ToType.

Definition qsort := Type.
(* qsort@{α ; u |} = Type@{α ; u} : Univ@{u+1} *)

Definition qsort' := Type.
(* qsort'@{u} : Type@{u+1} *)

Inductive implicit := mk.
(* implicit@{α ; u} : Type@{α ; u} *)

Inductive foo1 : Type := .
(* foo1@{α ; u |} : Type@{α ; _} :=  . *)

Inductive sigma (A:Type) (B:A -> Type) : Type
    := pair : forall x : A, B x -> sigma A B.
  (* sigma@{α α0 α1 ; u u0 u1} : forall (A : Type@{α ; u}) (_ : forall _ : A, Type@{α0 ; u0}), Type@{α1 ; u1} *)

(* Elimination constraints are added *)
Definition pr1 {A B} (s:sigma A B) : A
  := match s with pair _ _ x _ => x end.
(* α α0 α1 ; u u0 u1 |= α1 -> α *)

Definition pr2 {A B} (s:sigma A B) : B (pr1 s)
  := match s with pair _ _ _ y => y end.
(* α α0 α1 ; u u0 u1 |= α1 -> α
                                    α1 -> α0 *)

Overlays

Adapt to rocq-prover/rocq#21450 (Update esearch_fix_guard to return evd) mattam82/Coq-Equations#688

Added / updated test-suite.
Added changelog.
Added / updated documentation.
- Documented any new / changed user messages.
Opened overlay pull requests.

tabareau · 2026-01-06T16:56:38Z

@coqbot run full ci

vernac/declare.ml

SkySkimmer · 2026-02-17T14:09:54Z

@coqbot run full ci

SkySkimmer

I think if you squash everything into 1 commit this should be ready

TDiazT · 2026-02-19T12:27:58Z

@SkySkimmer squashing ready 👌

SkySkimmer · 2026-02-19T12:47:52Z

@coqbot run full ci

yannl35133 · 2026-02-19T18:55:40Z

kernel/inductive.ml


-
-let check_fix ?evars ?elim_to env ((nvect,_),(names,_,bodies as recdef) as fix) =
+(* Returns the pairs of (inductive sort * output sort) or None if Type in Type *)


Suggested change

(* Returns the pairs of (inductive sort * output sort) or None if Type in Type *)

(* Returns the pairs of (inductive sort * output sort)

or None if any elimination constraint was ignored. *)

yannl35133 · 2026-02-19T19:05:04Z

kernel/inductive.ml

+let sorts_of_mutfix env minds names =
+  let ind_ignores_elim_constraints (ind, _) = Environ.ind_ignores_elim_constraints env ind in
+  (* recursive sprop means non record with projections -> squashed *)
+  if Array.exists ind_ignores_elim_constraints minds then None


Why skip all constraints if only one is allowed to be ignored? If universe constraints are to be taken as a reference, we should keep all constraints, remembering whether they may be ignored, and try to add them all, only ignoring in case of failure.

This used to be a check of type in type, which made it unnecessary to check all cases. I assumed that the change to use ind_ignores_elim_constraints instead was including this type in type scenario, and so the logic remained the same. Is there more to this flag?

The currently proposed behaviour is fine, this can be modified in a later PR.

yannl35133 · 2026-02-19T19:07:48Z

pretyping/typing.ml

           uj_type = expected_type }

+let check_fix_with_elims env sigma fix =
+  let evars = Evd.evar_handler sigma in


As was suggested in another comment, this function should take in econstrs and deal with the translation to constr here : a hidden invariant is that check_fix_pre_sorts (because of check_one_fix) needs to take in evar-normalised terms, so that call to EConstr.to_constr should be reinstated (here instead).

doesn't check_fix_pre_sorts take a evar handler? why would it need evar normal terms?

check_one_fix doesn't evar-reduce before inspecting its body, so it may find an evar that is actually defined and not check the definition. The evar handler is only passed to the weak-head reducing function, which is not systematically called (for good reasons).

check_fix_sorts operating on constr can be changed at a later time, only EConstr.to_constr instead of Unsafe.to_constr is required.

yannl35133 · 2026-02-19T19:17:22Z

vernac/comInductive.ml

+  (* Or should inductive_levels be cut off earlier, e.g. at L646 ? *)
+  let candidates = if not @@ PolyFlags.collapse_sort_variables poly
+                   then []
+                   else prop_lowering_candidates evd ~arities_explicit inds


yannl35133 · 2026-02-19T19:18:23Z

vernac/declare.ml

+     [declare_mutual_definitions] where we get it from UState. *)
+  let sigma = Option.default (Evd.from_env env) sigma in
+  let res = Pretyping.search_guard env sigma possible_guard rec_declaration in
+  (* We are forgetting about sigma here -- probably missing something *)


This does look fishy, but I'm already surprised by the fact that this code handles regular constrs and not econstrs.

isn't it just an outdated comment?

Right, the evar_map isn't forgotten, but it's still weird that this code doesn't handle econstrs. This is unrelated to this PR now though.

@SkySkimmer

Currently behind a Sort Polymorphism flag. fix: Generate proper sort variable in record feat: Add sort elab for interactive definitions fix: Thread sort poly flag more correctly It was being passed as unconstrained_sorts in some cases. refactor: Remove unnecessary sort_poly flags fix: Add sort_poly to prepare_obligations refactor: Remove remaining fixmes and todos refactor: General cleanup refactor: Get ?evar_handler from evd refactor: Remove search_fix_guard doc: comment collapse refactor: Remove unnecessary unconstrained_sorts refactor: Remove unnecessary check for collapse_sort_variables revert: Some changes in unify quality feat: Add elim constraints to non-prim records with sort variables doc: Add doc on flag + implicit elab of sorts chore: Add changelog entry chore: Add overlays refactor: indentation feat: Add more specific elim constraints to non-prim projections Address comments refactor: Pass on sigma in search_guard test: Move sort_poly_elab test to output folder fix: Change assert implicit sort vars for error in sections fix(attempt): Return evd from make_recursive_bodies refactor: Change option array to list option refactor: Use Cmap_env to keep elim constraints from projs Apply suggestion from @SkySkimmer update test

yannl35133 · 2026-02-20T10:47:44Z

@coqbot run full ci

SkySkimmer · 2026-02-20T14:50:26Z

@coqbot merge now

coqbot-app · 2026-02-20T14:50:32Z

@SkySkimmer: Please take care of the following overlays:

21450-TDiazT-elab-sorts.sh

coqbot-app bot added the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Dec 22, 2025

TDiazT force-pushed the elab-sorts branch 3 times, most recently from cda08c0 to 26bf881 Compare December 24, 2025 13:19

github-actions bot added the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Dec 31, 2025

TDiazT force-pushed the elab-sorts branch from 26bf881 to ad55557 Compare January 6, 2026 14:53

coqbot-app bot removed the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Jan 6, 2026

coqbot-app bot removed the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Jan 6, 2026

TDiazT added a commit to TDiazT/rocq-lean-import that referenced this pull request Jan 7, 2026

Adapt to rocq-prover/rocq#21450 (Add poly flag to declare_projections)

e4d5676

TDiazT added a commit to TDiazT/Coq-Equations that referenced this pull request Jan 7, 2026

Adapt to rocq-prover/rocq#21450 (Update esearch_fix_guard to return evd)

a70592f

TDiazT added a commit to TDiazT/Coq-HoTT that referenced this pull request Jan 7, 2026

Adapt to rocq-prover/rocq#21450 (Section Univ -> Section Universe)

8d13271

TDiazT added a commit to TDiazT/coq-elpi that referenced this pull request Jan 7, 2026

Adapt to rocq-prover/rocq#21450 (Add poly flags to declare_projections)

a2000cf

github-actions bot added the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Jan 8, 2026

TDiazT force-pushed the elab-sorts branch from c138298 to 67189c9 Compare January 15, 2026 23:30

coqbot-app bot added needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. and removed needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. labels Jan 15, 2026

github-actions bot added the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Jan 16, 2026

TDiazT force-pushed the elab-sorts branch from 67189c9 to 9eed7dc Compare January 17, 2026 17:58

coqbot-app bot removed the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Jan 17, 2026

TDiazT force-pushed the elab-sorts branch 2 times, most recently from 4cfacbd to 3f8c2f6 Compare January 19, 2026 10:40

TDiazT marked this pull request as ready for review January 19, 2026 10:41

TDiazT requested review from a team as code owners January 19, 2026 10:41

TDiazT force-pushed the elab-sorts branch 2 times, most recently from c1101c4 to cc003ef Compare February 10, 2026 10:36

github-actions bot added the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Feb 11, 2026

TDiazT force-pushed the elab-sorts branch from cc003ef to 2b53439 Compare February 16, 2026 13:24

coqbot-app bot removed the needs: rebase Should be rebased on the latest master to solve conflicts or have a newer CI run. label Feb 16, 2026

SkySkimmer reviewed Feb 17, 2026

View reviewed changes

vernac/declare.ml Outdated Show resolved Hide resolved

coqbot-app bot removed the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Feb 17, 2026

SkySkimmer force-pushed the elab-sorts branch from 4edbda7 to 1eb5379 Compare February 18, 2026 12:24

coqbot-app bot removed the request: full CI Use this label when you want your next push to trigger a full CI. label Feb 18, 2026

SkySkimmer removed the request for review from a team February 18, 2026 12:25

SkySkimmer approved these changes Feb 18, 2026

View reviewed changes

TDiazT force-pushed the elab-sorts branch from 1eb5379 to c20a6dd Compare February 18, 2026 16:03

coqbot-app bot added the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Feb 18, 2026

coqbot-app bot removed the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Feb 19, 2026

SkySkimmer removed the needs: squashing Some commits should be squashed together. label Feb 19, 2026

yannl35133 reviewed Feb 19, 2026

View reviewed changes

TDiazT force-pushed the elab-sorts branch from c20a6dd to 167aa27 Compare February 20, 2026 10:44

coqbot-app bot added the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Feb 20, 2026

coqbot-app bot removed the needs: full CI The latest GitLab pipeline that ran was a light CI. Say "@coqbot run full ci" to get a full CI. label Feb 20, 2026



		let check_fix ?evars ?elim_to env ((nvect,_),(names,_,bodies as recdef) as fix) =
		(* Returns the pairs of (inductive sort * output sort) or None if Type in Type *)

	(* Returns the pairs of (inductive sort * output sort) or None if Type in Type *)
	(* Returns the pairs of (inductive sort * output sort)
	or None if any elimination constraint was ignored. *)

Conversation

TDiazT commented Dec 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Examples

Overlays

Uh oh!

tabareau commented Jan 6, 2026

Uh oh!

Uh oh!

SkySkimmer commented Feb 17, 2026

Uh oh!

SkySkimmer left a comment

Choose a reason for hiding this comment

Uh oh!

TDiazT commented Feb 19, 2026

Uh oh!

SkySkimmer commented Feb 19, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yannl35133 commented Feb 20, 2026

Uh oh!

SkySkimmer commented Feb 20, 2026

Uh oh!

coqbot-app bot commented Feb 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

TDiazT commented Dec 22, 2025 •

edited

Loading