We intend to provide regular “minor version” releases of the Lean language at approximately monthly intervals. There is not yet a strong guarantee of backwards compatibility between versions, only an expectation that breaking changes will be documented in this file.
This file contains work-in-progress notes for the upcoming release, as well as previous stable releases. Please check the releases page for the current status of each version.
Functions defined by well-founded recursion are now marked as @[irreducible], which should prevent expensive and often unfruitful unfolding of such definitions.
Existing proofs that hold by definitional equality (e.g. rfl) can be rewritten to explictly unfold the function definition (using simp, unfold, rw), or the recursive function can be temporariliy made semireducible (using unseal f in before the command) or the function definition itself can be marked as @[semireducible] to get the previous behavor.
Executables configured with supportInterpreter := true on Windows should now be run via lake exe to function properly.
The way Lean is built on Windows has changed (see PR #3601). As a result, Lake now dynamically links executables with supportInterpreter := true on Windows to libleanshared.dll and libInit_shared.dll. Therefore, such executables will not run unless those shared libraries are co-located with the executables or part of PATH. Running the executable via lake exe will ensure these libraries are part of PATH.
In a related change, the signature of the nativeFacets Lake configuration options has changed from a static Array to a function (shouldExport : Bool) → Array. See its docstring or Lake's README for further details on the changed option.
Lean now generates an error if the type of a theorem is not a proposition.
Importing two different files containing proofs of the same theorem is no longer considered an error. This feature is particularly useful for theorems that are automatically generated on demand (e.g., equational theorems).
Functional induction principles.
Derived from the definition of a (possibly mutually) recursive function, a functional induction principle is created that is tailored to proofs about that function.
For example from:
def ackermann : Nat → Nat → Nat | 0, m => m + 1 | n+1, 0 => ackermann n 1 | n+1, m+1 => ackermann n (ackermann (n + 1) m)
we get
ackermann.induct (motive : Nat → Nat → Prop) (case1 : ∀ (m : Nat), motive 0 m) (case2 : ∀ (n : Nat), motive n 1 → motive (Nat.succ n) 0) (case3 : ∀ (n m : Nat), motive (n + 1) m → motive n (ackermann (n + 1) m) → motive (Nat.succ n) (Nat.succ m)) (x x : Nat) : motive x x
It can be used in the induction tactic using the using syntax:
induction n, m using ackermann.induct
The termination checker now recognizes more recursion patterns without an explicit termination_by. In particular the idiom of counting up to an upper bound, as in
def Array.sum (arr : Array Nat) (i acc : Nat) : Nat :=
if _ : i < arr.size then
Array.sum arr (i+1) (acc + arr[i])
else
acc
is recognized without having to say termination_by arr.size - i.
Shorter instances names. There is a new algorithm for generating names for anonymous instances. Across Std and Mathlib, the median ratio between lengths of new names and of old names is about 72%. With the old algorithm, the longest name was 1660 characters, and now the longest name is 202 characters. The new algorithm's 95th percentile name length is 67 characters, versus 278 for the old algorithm. While the new algorithm produces names that are 1.2% less unique, it avoids cross-project collisions by adding a module-based suffix when it does not refer to declarations from the same “project” (modules that share the same root). PR #3089.
Attribute @[pp_using_anonymous_constructor] to make structures pretty print like ⟨x, y, z⟩ rather than {a := x, b := y, c := z}. This attribute is applied to Sigma, PSigma, PProd, Subtype, And, and Fin.
Now structure instances pretty print with parent structures' fields inlined. That is, if B extends A, then { toA := { x := 1 }, y := 2 } now pretty prints as { x := 1, y := 2 }. Setting option pp.structureInstances.flatten to false turns this off.
Option pp.structureProjections is renamed to pp.fieldNotation, and there is now a suboption pp.fieldNotation.generalized to enable pretty printing function applications using generalized field notation (defaults to true). Field notation can be disabled on a function-by-function basis using the @[pp_nodot] attribute.
Added options pp.mvars (default: true) and pp.mvars.withType (default: false). When pp.mvars is false, expression metavariables pretty print as ?_ and universe metavariables pretty print as _. When pp.mvars.withType is true, expression metavariables pretty print with a type ascription. These can be set when using #guard_msgs to make tests not depend on the particular names of metavariables. #3798 and #3978.
Hovers for terms in match expressions in the Infoview now reliably show the correct term.
Added @[induction_eliminator] and @[cases_eliminator] attributes to be able to define custom eliminators for the induction and cases tactics, replacing the @[eliminator] attribute. Gives custom eliminators for Nat so that induction and cases put goal states into terms of 0 and n + 1 rather than Nat.zero and Nat.succ n. Added option tactic.customEliminators to control whether to use custom eliminators. Added a hack for rcases/rintro/obtain to use the custom eliminator for Nat. #3629, #3655, and #3747.
The #guard_msgs command now has options to change whitespace normalization and sensitivity to message ordering. For example, #guard_msgs (whitespace := lax) in cmd collapses whitespace before checking messages, and #guard_msgs (ordering := sorted) in cmd sorts the messages in lexicographic order before checking. PR #3883.
The #guard_msgs command now supports showing a diff between the expected and actual outputs. This feature is currently disabled by default, but can be enabled with set_option guard_msgs.diff true. Depending on user feedback, this option may default to true in a future version of Lean.
Breaking changes:
.eq_<idx> instead of ._eq_<idx>, and .def instead of ._unfold. Example:def fact : Nat → Nat
| 0 => 1
| n+1 => (n+1) * fact n
theorem ex : fact 0 = 1 := by unfold fact; decide
#check fact.eq_1
-- fact.eq_1 : fact 0 = 1
#check fact.eq_2
-- fact.eq_2 (n : Nat) : fact (Nat.succ n) = (n + 1) * fact n
#check fact.def
/-
fact.def :
∀ (x : Nat),
fact x =
match x with
| 0 => 1
| Nat.succ n => (n + 1) * fact n
-/
The coercion from String to Name was removed. Previously, it was Name.mkSimple, which does not separate strings at dots, but experience showed that this is not always the desired coercion. For the previous behavior, manually insert a call to Name.mkSimple.
The Subarray fields as, h₁ and h₂ have been renamed to array, start_le_stop, and stop_le_array_size, respectively. This more closely follows standard Lean conventions. Deprecated aliases for the field projections were added; these will be removed in a future release.
The change to the instance name algorithm (described above) can break projects that made use of the auto-generated names.
Option.toMonad has been renamed to Option.getM and the unneeded [Monad m] instance argument has been removed.
simp and rw now use instance arguments found by unification, rather than always resynthesizing. For backwards compatibility, the original behaviour is available via set_option tactic.skipAssignedInstances false. #3507 and #3509.
When the pp.proofs is false, now omitted proofs use ⋯ rather than _, which gives a more helpful error message when copied from the Infoview. The pp.proofs.threshold option lets small proofs always be pretty printed. #3241.
pp.proofs.withType is now set to false by default to reduce noise in the info view.
The pretty printer for applications now handles the case of over-application itself when applying app unexpanders. In particular, the | `($_ $a $b $xs*) => `(($a + $b) $xs*) case of an app_unexpander is no longer necessary. #3495.
New simp (and dsimp) configuration option: zetaDelta. It is false by default. The zeta option is still true by default, but their meaning has changed.
zeta := true, simp and dsimp reduce terms of the form let x := val; e[x] into e[val].zetaDelta := true, simp and dsimp will expand let-variables in the context. For example, suppose the context contains x := val. Then, any occurrence of x is replaced with val.See issue #2682 for additional details. Here are some examples:
example (h : z = 9) : let x := 5; let y := 4; x + y = z := by
intro x
simp
/-
New goal:
h : z = 9; x := 5 |- x + 4 = z
-/
rw [h]
example (h : z = 9) : let x := 5; let y := 4; x + y = z := by
intro x
-- Using both `zeta` and `zetaDelta`.
simp (config := { zetaDelta := true })
/-
New goal:
h : z = 9; x := 5 |- 9 = z
-/
rw [h]
example (h : z = 9) : let x := 5; let y := 4; x + y = z := by
intro x
simp [x] -- asks `simp` to unfold `x`
/-
New goal:
h : z = 9; x := 5 |- 9 = z
-/
rw [h]
example (h : z = 9) : let x := 5; let y := 4; x + y = z := by
intro x
simp (config := { zetaDelta := true, zeta := false })
/-
New goal:
h : z = 9; x := 5 |- let y := 4; 5 + y = z
-/
rw [h]
When adding new local theorems to simp, the system assumes that the function application arguments have been annotated with no_index. This modification, which addresses issue #2670, restores the Lean 3 behavior that users expect. With this modification, the following examples are now operational:
example {α β : Type} {f : α × β → β → β} (h : ∀ p : α × β, f p p.2 = p.2)
(a : α) (b : β) : f (a, b) b = b := by
simp [h]
example {α β : Type} {f : α × β → β → β}
(a : α) (b : β) (h : f (a,b) (a,b).2 = (a,b).2) : f (a, b) b = b := by
simp [h]
In both cases, h is applicable because simp does not index f-arguments anymore when adding h to the simp-set. It's important to note, however, that global theorems continue to be indexed in the usual manner.
Improved the error messages produced by the decide tactic. #3422
Improved auto-completion performance. #3460
Improved initial language server startup performance. #3552
Changed call hierarchy to sort entries and strip private header from names displayed in the call hierarchy. #3482
There is now a low-level error recovery combinator in the parsing framework, primarily intended for DSLs. #3413
You can now write termination_by? after a declaration to see the automatically inferred termination argument, and turn it into a termination_by … clause using the “Try this” widget or a code action. #3514
A large fraction of Std has been moved into the Lean repository. This was motivated by:
ext, by_cases, change at, norm_cast, rcases, simpa, simp?, omega, and exact? available to all users of Lean, without imports.import Std.Nat, Int, Fin (and variants such as UInt64), List, Array, and BitVec as we begin making the APIs and simp normal forms for these types more complete and consistent.RBMap) and utilities such as basic IO. While we have achieved most of our initial aims in v4.7.0-rc1, some upstreaming will continue over the coming months.The / and % notations in Int now use Int.ediv and Int.emod (i.e. the rounding conventions have changed). Previously Std overrode these notations, so this is no change for users of Std. There is now kernel support for these functions. #3376.
omega, our integer linear arithmetic tactic, is now availabe in the core langauge.
bv_omega which can solve goals about BitVec which naturally translate into linear arithmetic problems. #3435.omega now has support for Fin #3427, the <<< operator #3433.omega was modified to no longer identify atoms up to definitional equality (so in particular it can no longer prove id x ≤ x). #3525. This may cause some regressions. We plan to provide a general purpose preprocessing tactic later, or an omega! mode.omega is now invoked in Lean's automation for termination proofs #3503 as well as in array indexing proofs #3515. This automation will be substantially revised in the medium term, and while omega does help automate some proofs, we plan to make this much more robust.The library search tactics exact? and apply? that were originally in Mathlib are now available in Lean itself. These use the implementation using lazy discrimination trees from Std, and thus do not require a disk cache but have a slightly longer startup time. The order used for selection lemmas has changed as well to favor goals purely based on how many terms in the head pattern match the current goal.
The solve_by_elim tactic has been ported from Std to Lean so that library search can use it.
New #check_tactic and #check_simp commands have been added. These are useful for checking tactics (particularly simp) behave as expected in test suites.
Previously, app unexpanders would only be applied to entire applications. However, some notations produce functions, and these functions can be given additional arguments. The solution so far has been to write app unexpanders so that they can take an arbitrary number of additional arguments. However this leads to misleading hover information in the Infoview. For example, while HAdd.hAdd f g 1 pretty prints as (f + g) 1, hovering over f + g shows f. There is no way to fix the situation from within an app unexpander; the expression position for HAdd.hAdd f g is absent, and app unexpanders cannot register TermInfo.
This commit changes the app delaborator to try running app unexpanders on every prefix of an application, from longest to shortest prefix. For efficiency, it is careful to only try this when app delaborators do in fact exist for the head constant, and it also ensures arguments are only delaborated once. Then, in (f + g) 1, the f + g gets TermInfo registered for that subexpression, making it properly hoverable.
Breaking changes:
Lean.withTraceNode and variants got a stronger MonadAlwaysExcept assumption to fix trace trees not being built on elaboration runtime exceptions. Instances for most elaboration monads built on EIO Exception should be synthesized automatically.match ... with. and fun. notations previously in Std have been replaced by nomatch ... and nofun. #3279 and #3286Other improvements:
simp:match expressions:termination_by error messages #3255rename_i in macros, fixes #3553 #3581generalize, fixes #3524 #3575add_decl_doc should check that declarations are local #3311Lake fixes:
Add custom simplification procedures (aka simprocs) to simp. Simprocs can be triggered by the simplifier on a specified term-pattern. Here is an small example:
import Lean.Meta.Tactic.Simp.BuiltinSimprocs.Nat def foo (x : Nat) : Nat := x + 10 /-- The `simproc` `reduceFoo` is invoked on terms that match the pattern `foo _`. -/ simproc reduceFoo (foo _) := /- A term of type `Expr → SimpM Step -/ fun e => do /- The `Step` type has three constructors: `.done`, `.visit`, `.continue`. * The constructor `.done` instructs `simp` that the result does not need to be simplied further. * The constructor `.visit` instructs `simp` to visit the resulting expression. * The constructor `.continue` instructs `simp` to try other simplification procedures. All three constructors take a `Result`. The `.continue` contructor may also take `none`. `Result` has two fields `expr` (the new expression), and `proof?` (an optional proof). If the new expression is definitionally equal to the input one, then `proof?` can be omitted or set to `none`. -/ /- `simp` uses matching modulo reducibility. So, we ensure the term is a `foo`-application. -/ unless e.isAppOfArity ``foo 1 do return .continue /- `Nat.fromExpr?` tries to convert an expression into a `Nat` value -/ let some n ← Nat.fromExpr? e.appArg! | return .continue return .done { expr := Lean.mkNatLit (n+10) }
We disable simprocs support by using the command set_option simprocs false. This command is particularly useful when porting files to v4.6.0. Simprocs can be scoped, manually added to simp commands, and suppressed using -. They are also supported by simp?. simp only does not execute any simproc. Here are some examples for the simproc defined above.
example : x + foo 2 = 12 + x := by set_option simprocs false in /- This `simp` command does not make progress since `simproc`s are disabled. -/ fail_if_success simp simp_arith example : x + foo 2 = 12 + x := by /- `simp only` must not use the default simproc set. -/ fail_if_success simp only simp_arith example : x + foo 2 = 12 + x := by /- `simp only` does not use the default simproc set, but we can provide simprocs as arguments. -/ simp only [reduceFoo] simp_arith example : x + foo 2 = 12 + x := by /- We can use `-` to disable `simproc`s. -/ fail_if_success simp [-reduceFoo] simp_arith
The command register_simp_attr <id> now creates a simp and a simproc set with the name <id>. The following command instructs Lean to insert the reduceFoo simplification procedure into the set my_simp. If no set is specified, Lean uses the default simp set.
simproc [my_simp] reduceFoo (foo _) := ...
The syntax of the termination_by and decreasing_by termination hints is overhauled:
mutual block.where clause, the termination_by and decreasing_by for that function come before the where. The functions in the where clause can have their own termination hints, each following the corresponding definition.termination_by clause can only bind “extra parameters”, that are not already bound by the function header, but are bound in a lambda (:= fun x y z =>) or in patterns (| x, n + 1 => …). These extra parameters used to be understood as a suffix of the function parameters; now it is a prefix.Migration guide: In simple cases just remove the function name, and any variables already bound at the header.
def foo : Nat → Nat → Nat := … -termination_by foo a b => a - b +termination_by a b => a - b
or
def foo : Nat → Nat → Nat := … -termination_by _ a b => a - b +termination_by a b => a - b
If the parameters are bound in the function header (before the :), remove them as well:
def foo (a b : Nat) : Nat := … -termination_by foo a b => a - b +termination_by a - b
Else, if there are multiple extra parameters, make sure to refer to the right ones; the bound variables are interpreted from left to right, no longer from right to left:
def foo : Nat → Nat → Nat → Nat | a, b, c => … -termination_by foo b c => b +termination_by a b => b
In the case of a mutual block, place the termination arguments (without the function name) next to the function definition:
-mutual -def foo : Nat → Nat → Nat := … -def bar : Nat → Nat := … -end -termination_by - foo a b => a - b - bar a => a +mutual +def foo : Nat → Nat → Nat := … +termination_by a b => a - b +def bar : Nat → Nat := … +termination_by a => a +end
Similarly, if you have (mutual) recursion through where or let rec, the termination hints are now placed directly after the function they apply to:
-def foo (a b : Nat) : Nat := … - where bar (x : Nat) : Nat := … -termination_by - foo a b => a - b - bar x => x +def foo (a b : Nat) : Nat := … +termination_by a - b + where + bar (x : Nat) : Nat := … + termination_by x -def foo (a b : Nat) : Nat := - let rec bar (x : Nat) : Nat := … - … -termination_by - foo a b => a - b - bar x => x +def foo (a b : Nat) : Nat := + let rec bar (x : Nat) : Nat := … + termination_by x + … +termination_by a - b
In cases where a single decreasing_by clause applied to multiple mutually recursive functions before, the tactic now has to be duplicated.
The semantics of decreasing_by changed; the tactic is applied to all termination proof goals together, not individually.
This helps when writing termination proofs interactively, as one can focus each subgoal individually, for example using ·. Previously, the given tactic script had to work for all goals, and one had to resort to tactic combinators like first:
def foo (n : Nat) := … foo e1 … foo e2 … -decreasing_by -simp_wf -first | apply something_about_e1; … - | apply something_about_e2; … +decreasing_by +all_goals simp_wf +· apply something_about_e1; … +· apply something_about_e2; …
To obtain the old behaviour of applying a tactic to each goal individually, use all_goals:
def foo (n : Nat) := … -decreasing_by some_tactic +decreasing_by all_goals some_tactic
In the case of mutual recursion each decreasing_by now applies to just its function. If some functions in a recursive group do not have their own decreasing_by, the default decreasing_tactic is used. If the same tactic ought to be applied to multiple functions, the decreasing_by clause has to be repeated at each of these functions.
Modify InfoTree.context to facilitate augmenting it with partial contexts while elaborating a command. This breaks backwards compatibility with all downstream projects that traverse the InfoTree manually instead of going through the functions in InfoUtils.lean, as well as those manually creating and saving InfoTrees. See PR #3159 for how to migrate your code.
Add language server support for call hierarchy requests (PR #3082). The change to the .ilean format in this PR means that projects must be fully rebuilt once in order to generate .ilean files with the new format before features like “find references” work correctly again.
Structure instances with multiple sources (for example {a, b, c with x := 0}) now have their fields filled from these sources in strict left-to-right order. Furthermore, the structure instance elaborator now aggressively use sources to fill in subobject fields, which prevents unnecessary eta expansion of the sources, and hence greatly reduces the reliance on costly structure eta reduction. This has a large impact on mathlib, reducing total CPU instructions by 3% and enabling impactful refactors like leanprover-community/mathlib4#8386 which reduces the build time by almost 20%. See PR #2478 and RFC #2451.
Add pretty printer settings to omit deeply nested terms (pp.deepTerms false and pp.deepTerms.threshold) (PR #3201)
Add pretty printer options pp.numeralTypes and pp.natLit. When pp.numeralTypes is true, then natural number literals, integer literals, and rational number literals are pretty printed with type ascriptions, such as (2 : Rat), (-2 : Rat), and (-2 / 3 : Rat). When pp.natLit is true, then raw natural number literals are pretty printed as nat_lit 2. PR #2933 and RFC #3021.
Lake updates:
Other improvements:
intro be aware of let_fun #3115rw #3121mkCongrArg calls in proofs generated by simp #3203induction using followed by a general term #3188let #3060, fixing #3065swap! should return a, not `default`` #3197, fixing #3196BEq on structure with Prop-fields #3191, fixing #3140casesOnApp/matcherApp #3176, fixing #3175deriving only deriving the first declaration for some handlers #3058, fixing #3057cases h : ... #3084Modify the lexical syntax of string literals to have string gaps, which are escape sequences of the form "\" newline whitespace*. These have the interpetation of an empty string and allow a string to flow across multiple lines without introducing additional whitespace. The following is equivalent to "this is a string".
"this is \ a string"
Add raw string literal syntax. For example, r"\n" is equivalent to "\\n", with no escape processing. To include double quote characters in a raw string one can add sufficiently many # characters before and after the bounding "s, as in r#"the "the" is in quotes"# for "the \"the\" is in quotes". PR #2929 and issue #1422.
The low-level termination_by' clause is no longer supported.
Migration guide: Use termination_by instead, e.g.:
-termination_by' measure (fun ⟨i, _⟩ => as.size - i) +termination_by i _ => as.size - i
If the well-founded relation you want to use is not the one that the WellFoundedRelation type class would infer for your termination argument, you can use WellFounded.wrap from the std libarary to explicitly give one:
-termination_by' ⟨r, hwf⟩ +termination_by x => hwf.wrap x
Support snippet edits in LSP TextEdits. See Lean.Lsp.SnippetString for more details.
Deprecations and changes in the widget API.
Widget.UserWidgetDefinition is deprecated in favour of Widget.Module. The annotation @[widget] is deprecated in favour of @[widget_module]. To migrate a definition of type UserWidgetDefinition, remove the name field and replace the type with Widget.Module. Removing the name results in a title bar no longer being drawn above your panel widget. To add it back, draw it as part of the component using <details open=true><summary class='mv2 pointer'>{name}</summary>{rest_of_widget}</details>. See an example migration here.show_panel_widgets allows displaying always-on and locally-on panel widgets.RpcEncodable widget props can now be stored in the infotree.If no usable lexicographic order can be found automatically for a termination proof, explain why. See feat: GuessLex: if no measure is found, explain why.
Option to print inferred termination argument. With set_option showInferredTerminationBy true you will get messages like
Inferred termination argument: termination_by ackermann n m => (sizeOf n, sizeOf m)
for automatically generated termination_by clauses.
More detailed error messages for invalid mutual blocks.
Multiple improvements to the output of simp? and simp_all?.
Tactics with withLocation * no longer fail if they close the main goal.
Implementation of a test_extern command for writing tests for @[extern] and @[implemented_by] functions. Usage is
import Lean.Util.TestExtern test_extern Nat.add 17 37
The head symbol must be the constant with the @[extern] or @[implemented_by] attribute. The return type must have a DecidableEq instance.
Bug fixes for #2853, #2953, #2966, #2971, #2990, #3094.
Bug fix for eager evaluation of default value in Option.getD. Avoid panic in leanPosToLspPos when file source is unavailable. Improve short-circuiting behavior for List.all and List.any.
Several Lake bug fixes: #3036, #3064, #3069.
Lake and the language server now support per-package server options using the moreServerOptions config field, as well as options that apply to both the language server and lean using the leanOptions config field. Setting either of these fields instead of moreServerArgs ensures that viewing files from a dependency uses the options for that dependency. Additionally, moreServerArgs is being deprecated in favor of the moreGlobalServerArgs field. See PR #2858.
A Lakefile with the following deprecated package declaration:
def moreServerArgs := #[ "-Dpp.unicode.fun=true" ] def moreLeanArgs := moreServerArgs package SomePackage where moreServerArgs := moreServerArgs moreLeanArgs := moreLeanArgs
... can be updated to the following package declaration to use per-package options:
package SomePackage where leanOptions := #[⟨`pp.unicode.fun, true⟩]
Bug fixes for #2628, #2883, #2810, #2925, and #2914.
Lake:
lake init . and a bare lake init and will now use the current directory as the package name. #2890lake new and lake init will now produce errors on invalid package names such as .., foo/bar, Init, Lean, Lake, and Main. See issue #2637 and PR #2890.lean_lib no longer converts its name to upper camel case (e.g., lean_lib bar will include modules named bar.* rather than Bar.*). See issue #2567 and PR #2889.lake new 123-hello and import «123Hello» now work correctly). See issue #2865 and PR #2889.lakefile.olean) to include only those relevant to Lake's workspace loading process. This resolves segmentation faults caused by environment extension type mismatches (e.g., when defining custom elaborators via elab in configurations). See issue #2632 and PR #2896.math template has been simplified. See PR #2930.lake exe <target> now parses target like a build target (as the help text states it should) rather than as a basic name. For example, lake exe @mathlib/runLinter should now work. See PR #2932.lake new foo.bar [std] now generates executables named foo-bar and lake new foo.bar exe properly creates foo/bar.lean. See PR #2932.findModule?. See PR #2937.simp [f] does not unfold partial applications of f anymore. See issue #2042. To fix proofs affected by this change, use unfold f or simp (config := { unfoldPartialApp := true }) [f].
By default, simp will no longer try to use Decidable instances to rewrite terms. In particular, not all decidable goals will be closed by simp, and the decide tactic may be useful in such cases. The decide simp configuration option can be used to locally restore the old simp behavior, as in simp (config := {decide := true}); this includes using Decidable instances to verify side goals such as numeric inequalities.
Many bug fixes:
Decidable instances very slow when using cases tacticsimp not rewriting in bindersimp unfolding let even with zeta := false optionsimp (with beta/zeta disabled) and discrimination treesrw ... at hdsimp doesn't use rfl theorems which consist of an unapplied constantdsimp does not close reflexive equality goals if they are wrapped in metadatarw [h] uses h from the environment in preference to h from the local contextwithAssignableSyntheticOpaque for assumption tacticCancel outstanding tasks on document edit in the language server.
Lake:
lake new MyProject mathpostUpdate? configuration option to a post_update declaration. See the post_update syntax docstring for more information on the new syntax.:= syntax for configuration declarations (i.e., package, lean_lib, and lean_exe) has been deprecated. For example, package foo := {...} is deprecated.LAKE_PKG_URL_MAPbuild), default packages directory (e.g., lake-packages), and the compiled configuration (e.g., lakefile.olean) into a new dedicated directory for Lake outputs, .lake. The cloud release build archives are also stored here, fixing #2713.manifestFile field of a package configuration.lakefile.olean compatibility (see #2842 for more details).Environment.mk and Environment.add private, and add replay as a safer alternative.IO.Process.output no longer inherits the standard input of the caller.match reduction.DecidableEq now handles mutual inductive types.postUpdate? package configuration option. Used by a package to specify some code which should be run after a successful lake update of the package or one of its downstream dependencies. (lake#185)refine e now replaces the main goal with metavariables which were created during elaboration of e and no longer captures pre-existing metavariables that occur in e (#2502).withCollectingNewGoalsFrom, which also affects elabTermWithHoles, refine', calc (tactic), and specialize. Likewise, all of these now only include newly-created metavariables in their output.e were returned inconsistently in different edge cases, causing duplicated goals in the infoview (issue #2495), erroneously closed goals (issue #2434), and unintuitive behavior due to refine e capturing previously-created goals appearing unexpectedly in e (no issue; see PR).The error positioning on missing tokens has been improved. In particular, this should make it easier to spot errors in incomplete tactic proofs.
After elaborating a configuration file, Lake will now cache the configuration to a lakefile.olean. Subsequent runs of Lake will import this OLean instead of elaborating the configuration file. This provides a significant performance improvement (benchmarks indicate that using the OLean cuts Lake's startup time in half), but there are some important details to keep in mind:
lakefile.lean or lean-toolchain. You can also force a reconfigure by passing the new --reconfigure / -R option to lake.-K) will be fixed at the moment of elaboration. Setting these options when lake is using the cached configuration will have no effect. To change options, run lake with -R / --reconfigure.lakefile.olean is a local configuration and should not be committed to Git. Therefore, existing Lake packages need to add it to their .gitignore.The signature of Lake.buildO has changed, args has been split into weakArgs and traceArgs. traceArgs are included in the input trace and weakArgs are not. See Lake's FFI example for a demonstration of how to adapt to this change.
The signatures of Lean.importModules, Lean.Elab.headerToImports, and Lean.Elab.parseImports have changed from taking List Import to Array Import.
There is now an occs field in the configuration object for the rewrite tactic, allowing control of which occurrences of a pattern should be rewritten. This was previously a separate argument for Lean.MVarId.rewrite, and this has been removed in favour of an additional field of Rewrite.Config. It was not previously accessible from user tactics.
Lean.Meta.getConst? has been renamed. We have renamed getConst? to getUnfoldableConst? (and getConstNoEx? to getUnfoldableConstNoEx?). These were not intended to be part of the public API, but downstream projects had been using them (sometimes expecting different behaviour) incorrectly instead of Lean.getConstInfo.
dsimp / simp / simp_all now fail by default if they make no progress.
This can be overridden with the (config := { failIfUnchanged := false }) option. This change was made to ease manual use of simp (with complicated goals it can be hard to tell if it was effective) and to allow easier flow control in tactics internally using simp. See the summary discussion on zulip for more details.
simp_all now preserves order of hypotheses.
In order to support the failIfUnchanged configuration option for dsimp / simp / simp_all the way simp_all replaces hypotheses has changed. In particular it is now more likely to preserve the order of hypotheses. See simp_all reorders hypotheses unnecessarily. (Previously all non-dependent propositional hypotheses were reverted and reintroduced. Now only such hypotheses which were changed, or which come after a changed hypothesis, are reverted and reintroduced. This has the effect of preserving the ordering amongst the non-dependent propositional hypotheses, but now any dependent or non-propositional hypotheses retain their position amongst the unchanged non-dependent propositional hypotheses.) This may affect proofs that use rename_i, case ... =>, or next ... =>.
this is now a regular identifier again that is implicitly introduced by anonymous have := for the remainder of the tactic block. It used to be a keyword that was visible in all scopes and led to unexpected behavior when explicitly used as a binder name.
Make calc require the sequence of relation/proof-s to have the same indentation, and add calc alternative syntax allowing underscores _ in the first relation.
The flexible indentation in calc was often used to align the relation symbols:
example (x y : Nat) : (x + y) * (x + y) = x * x + y * x + x * y + y * y := calc (x + y) * (x + y) = (x + y) * x + (x + y) * y := by rw [Nat.mul_add] -- improper indentation _ = x * x + y * x + (x + y) * y := by rw [Nat.add_mul] _ = x * x + y * x + (x * y + y * y) := by rw [Nat.add_mul] _ = x * x + y * x + x * y + y * y := by rw [←Nat.add_assoc]
This is no longer legal. The new syntax puts the first term right after the calc and each step has the same indentation:
example (x y : Nat) : (x + y) * (x + y) = x * x + y * x + x * y + y * y := calc (x + y) * (x + y) _ = (x + y) * x + (x + y) * y := by rw [Nat.mul_add] _ = x * x + y * x + (x + y) * y := by rw [Nat.add_mul] _ = x * x + y * x + (x * y + y * y) := by rw [Nat.add_mul] _ = x * x + y * x + x * y + y * y := by rw [←Nat.add_assoc]
Update Lake to latest prerelease.
Make go-to-definition on a typeclass projection application go to the instance(s).
Add linter.deprecated option to silence deprecation warnings.
simp can track information and can print an equivalent simp only. PR #1626.
Enforce uniform indentation in tactic blocks / do blocks. See issue #1606.
Moved AssocList, HashMap, HashSet, RBMap, RBSet, PersistentArray, PersistentHashMap, PersistentHashSet to the Lean package. The standard library contains versions that will evolve independently to simplify bootstrapping process.
Standard library moved to the std4 GitHub repository.
InteractiveGoals now has information that a client infoview can use to show what parts of the goal have changed after applying a tactic. PR #1610.
Add [inheritDoc] attribute. PR #1480.
Expose that panic = default. PR #1614.
New code generator project has started.
Remove description argument from register_simp_attr. PR #1566.
Many new doc strings have been added to declarations at Init.
Update Lake to v4.0.0. See the v4.0.0 release notes for detailed changes.
Mutual declarations in different namespaces are now supported. Example:
mutual def Foo.boo (x : Nat) := match x with | 0 => 1 | x + 1 => 2*Boo.bla x def Boo.bla (x : Nat) := match x with | 0 => 2 | x+1 => 3*Foo.boo x end
A namespace is automatically created for the common prefix. Example:
mutual def Tst.Foo.boo (x : Nat) := ... def Tst.Boo.bla (x : Nat) := ... end
expands to
namespace Tst mutual def Foo.boo (x : Nat) := ... def Boo.bla (x : Nat) := ... end end Tst
Allow users to install their own deriving handlers for existing type classes. See example at Simple.lean.
Add tactic congr (num)?. See doc string for additional details.
match-syntax notation now checks for unused alternatives. See issue #1371.
Auto-completion for structure instance fields. Example:
example : Nat × Nat := { f -- HERE }
fst now appears in the list of auto-completion suggestions.
Auto-completion for dotted identifier notation. Example:
example : Nat := .su -- HERE
succ now appears in the list of auto-completion suggestions.
nat_lit is not needed anymore when declaring OfNat instances. See issues #1389 and #875. Example:
inductive Bit where | zero | one instance inst0 : OfNat Bit 0 where ofNat := Bit.zero instance : OfNat Bit 1 where ofNat := Bit.one example : Bit := 0 example : Bit := 1
Add [elabAsElim] attribute (it is called elab_as_eliminator in Lean 3). Motivation: simplify the Mathlib port to Lean 4.
Trans type class now accepts relations in Type u. See this Zulip issue.
Accept unescaped keywords as inductive constructor names. Escaping can often be avoided at use sites via dot notation.
inductive MyExpr | let : ... def f : MyExpr → MyExpr | .let ... => .let ...
Throw an error message at parametric local instances such as [Nat -> Decidable p]. The type class resolution procedure cannot use this kind of local instance because the parameter does not have a forward dependency. This check can be disabled using set_option checkBinderAnnotations false.
Add option pp.showLetValues. When set to false, the info view hides the value of let-variables in a goal. By default, it is true when visualizing tactic goals, and false otherwise. See issue #1345 for additional details.
Add option warningAsError. When set to true, warning messages are treated as errors.
Support dotted notation and named arguments in patterns. Example:
def getForallBinderType (e : Expr) : Expr := match e with | .forallE (binderType := type) .. => type | _ => panic! "forall expected"
“jump-to-definition” now works for function names embedded in the following attributes @[implementedBy funName], @[tactic parserName], @[termElab parserName], @[commandElab parserName], @[builtinTactic parserName], @[builtinTermElab parserName], and @[builtinCommandElab parserName]. See issue #1350.
Improve MVarId methods discoverability. See issue #1346. We still have to add similar methods for FVarId, LVarId, Expr, and other objects. Many existing methods have been marked as deprecated.
Add attribute [deprecated] for marking deprecated declarations. Examples:
def g (x : Nat) := x + 1 -- Whenever `f` is used, a warning message is generated suggesting to use `g` instead. @[deprecated g] def f (x : Nat) := x + 1 #check f 0 -- warning: `f` has been deprecated, use `g` instead -- Whenever `h` is used, a warning message is generated. @[deprecated] def h (x : Nat) := x + 1 #check h 0 -- warning: `h` has been deprecated
Add type LevelMVarId (and abbreviation LMVarId) for universe level metavariable ids. Motivation: prevent meta-programmers from mixing up universe and expression metavariable ids.
Improve calc term and tactic. See issue #1342.
Relaxed antiquotation parsing further reduces the need for explicit $x:p antiquotation kind annotations.
Add support for computed fields in inductives. Example:
inductive Exp | var (i : Nat) | app (a b : Exp) with @[computedField] hash : Exp → Nat | .var i => i | .app a b => a.hash * b.hash + 1
The result of the Exp.hash function is then stored as an extra “computed” field in the .var and .app constructors; Exp.hash accesses this field and thus runs in constant time (even on dag-like values).
Update a[i] notation. It is now based on the typeclass
class GetElem (cont : Type u) (idx : Type v) (elem : outParam (Type w)) (dom : outParam (cont → idx → Prop)) where getElem (xs : cont) (i : idx) (h : dom xs i) : Elem
The notation a[i] is now defined as follows
macro:max x:term noWs "[" i:term "]" : term => `(getElem $x $i (by get_elem_tactic))
The proof that i is a valid index is synthesized using the tactic get_elem_tactic. For example, the type Array α has the following instances
instance : GetElem (Array α) Nat α fun xs i => LT.lt i xs.size where ... instance : GetElem (Array α) USize α fun xs i => LT.lt i.toNat xs.size where ...
You can use the notation a[i]'h to provide the proof manually. Two other notations were introduced: a[i]! and a[i]?, For a[i]!, a panic error message is produced at runtime if i is not a valid index. a[i]? has type Option α, and a[i]? evaluates to none if the index i is not valid. The three new notations are defined as follows:
@[inline] def getElem' [GetElem cont idx elem dom] (xs : cont) (i : idx) (h : dom xs i) : elem := getElem xs i h @[inline] def getElem! [GetElem cont idx elem dom] [Inhabited elem] (xs : cont) (i : idx) [Decidable (dom xs i)] : elem := if h : _ then getElem xs i h else panic! "index out of bounds" @[inline] def getElem? [GetElem cont idx elem dom] (xs : cont) (i : idx) [Decidable (dom xs i)] : Option elem := if h : _ then some (getElem xs i h) else none macro:max x:term noWs "[" i:term "]" noWs "?" : term => `(getElem? $x $i) macro:max x:term noWs "[" i:term "]" noWs "!" : term => `(getElem! $x $i) macro x:term noWs "[" i:term "]'" h:term:max : term => `(getElem' $x $i $h)
See discussion on Zulip. Examples:
example (a : Array Int) (i : Nat) : Int := a[i] -- Error: failed to prove index is valid ... example (a : Array Int) (i : Nat) (h : i < a.size) : Int := a[i] -- Ok example (a : Array Int) (i : Nat) : Int := a[i]! -- Ok example (a : Array Int) (i : Nat) : Option Int := a[i]? -- Ok example (a : Array Int) (h : a.size = 2) : Int := a[0]'(by rw [h]; decide) -- Ok example (a : Array Int) (h : a.size = 2) : Int := have : 0 < a.size := by rw [h]; decide have : 1 < a.size := by rw [h]; decide a[0] + a[1] -- Ok example (a : Array Int) (i : USize) (h : i.toNat < a.size) : Int := a[i] -- Ok
The get_elem_tactic is defined as
macro "get_elem_tactic" : tactic => `(first | get_elem_tactic_trivial | fail "failed to prove index is valid, ..." )
The get_elem_tactic_trivial auxiliary tactic can be extended using macro_rules. By default, it tries trivial, simp_arith, and a special case for Fin. In the future, it will also try linarith. You can extend get_elem_tactic_trivial using my_tactic as follows
macro_rules | `(tactic| get_elem_tactic_trivial) => `(tactic| my_tactic)
Note that Idx's type in GetElem does not depend on Cont. So, you cannot write the instance instance : GetElem (Array α) (Fin ??) α fun xs i => ..., but the Lean library comes equipped with the following auxiliary instance:
instance [GetElem cont Nat elem dom] : GetElem cont (Fin n) elem fun xs i => dom xs i where getElem xs i h := getElem xs i.1 h
and helper tactic
macro_rules | `(tactic| get_elem_tactic_trivial) => `(tactic| apply Fin.val_lt_of_le; get_elem_tactic_trivial; done)
Example:
example (a : Array Nat) (i : Fin a.size) := a[i] -- Ok example (a : Array Nat) (h : n ≤ a.size) (i : Fin n) := a[i] -- Ok
Better support for qualified names in recursive declarations. The following is now supported:
namespace Nat def fact : Nat → Nat | 0 => 1 | n+1 => (n+1) * Nat.fact n end Nat
Add support for CommandElabM monad at #eval. Example:
import Lean open Lean Elab Command #eval do let id := mkIdent `foo elabCommand (← `(def $id := 10)) #eval foo -- 10
Try to elaborate do notation even if the expected type is not available. We still delay elaboration when the expected type is not available. This change is particularly useful when writing examples such as
#eval do IO.println "hello" IO.println "world"
That is, we don't have to use the idiom #eval show IO _ from do ... anymore. Note that auto monadic lifting is less effective when the expected type is not available. Monadic polymorphic functions (e.g., ST.Ref.get) also require the expected type.
On Linux, panics now print a backtrace by default, which can be disabled by setting the environment variable LEAN_BACKTRACE to 0. Other platforms are TBD.
The group(·) syntax combinator is now introduced automatically where necessary, such as when using multiple parsers inside (...)+.
Add “Typed Macros”: syntax trees produced and accepted by syntax antiquotations now remember their syntax kinds, preventing accidental production of ill-formed syntax trees and reducing the need for explicit :kind antiquotation annotations. See PR for details.
Aliases of protected definitions are protected too. Example:
protected def Nat.double (x : Nat) := 2*x namespace Ex export Nat (double) -- Add alias Ex.double for Nat.double end Ex open Ex #check Ex.double -- Ok #check double -- Error, `Ex.double` is alias for `Nat.double` which is protected
Use IO.getRandomBytes to initialize random seed for IO.rand. See discussion at this PR.
Improve dot notation and aliases interaction. See discussion on Zulip for additional details. Example:
def Set (α : Type) := α → Prop def Set.union (s₁ s₂ : Set α) : Set α := fun a => s₁ a ∨ s₂ a def FinSet (n : Nat) := Fin n → Prop namespace FinSet export Set (union) -- FinSet.union is now an alias for `Set.union` end FinSet example (x y : FinSet 10) : FinSet 10 := x.union y -- Works
ext and enter conv tactics can now go inside let-declarations. Example:
example (g : Nat → Nat) (y : Nat) (h : let x := y + 1; g (0+x) = x) : g (y + 1) = y + 1 := by conv at h => enter [x, 1, 1]; rw [Nat.zero_add] /- g : Nat → Nat y : Nat h : let x := y + 1; g x = x ⊢ g (y + 1) = y + 1 -/ exact h
Add zeta conv tactic to expand let-declarations. Example:
example (h : let x := y + 1; 0 + x = y) : False := by conv at h => zeta; rw [Nat.zero_add] /- y : Nat h : y + 1 = y ⊢ False -/ simp_arith at h
Improve namespace resolution. See issue #1224. Example:
import Lean open Lean Parser Elab open Tactic -- now opens both `Lean.Parser.Tactic` and `Lean.Elab.Tactic`
Rename constant command to opaque. See discussion at Zulip.
Extend induction and cases syntax: multiple left-hand-sides in a single alternative. This extension is very similar to the one implemented for match expressions. Examples:
inductive Foo where | mk1 (x : Nat) | mk2 (x : Nat) | mk3 def f (v : Foo) := match v with | .mk1 x => x + 1 | .mk2 x => 2*x + 1 | .mk3 => 1 theorem f_gt_zero : f v > 0 := by cases v with | mk1 x | mk2 x => simp_arith! -- New feature used here! | mk3 => decide
Add unnamed antiquotation $_ for use in syntax quotation patterns.
Add unused variables linter. Feedback welcome!
Lean now generates an error if the body of a declaration body contains a universe parameter that does not occur in the declaration type, nor is an explicit parameter. Examples:
/- The following declaration now produces an error because `PUnit` is universe polymorphic, but the universe parameter does not occur in the function type `Nat → Nat` -/ def f (n : Nat) : Nat := let aux (_ : PUnit) : Nat := n + 1 aux ⟨⟩ /- The following declaration is accepted because the universe parameter was explicitly provided in the function signature. -/ def g.{u} (n : Nat) : Nat := let aux (_ : PUnit.{u}) : Nat := n + 1 aux ⟨⟩
Add subst_vars tactic.
Fix autoParam in structure fields lost in multiple inheritance..
Add [eliminator] attribute. It allows users to specify default recursor/eliminators for the induction and cases tactics. It is an alternative for the using notation. Example:
@[eliminator] protected def recDiag {motive : Nat → Nat → Sort u} (zero_zero : motive 0 0) (succ_zero : (x : Nat) → motive x 0 → motive (x + 1) 0) (zero_succ : (y : Nat) → motive 0 y → motive 0 (y + 1)) (succ_succ : (x y : Nat) → motive x y → motive (x + 1) (y + 1)) (x y : Nat) : motive x y := let rec go : (x y : Nat) → motive x y | 0, 0 => zero_zero | x+1, 0 => succ_zero x (go x 0) | 0, y+1 => zero_succ y (go 0 y) | x+1, y+1 => succ_succ x y (go x y) go x y termination_by go x y => (x, y) def f (x y : Nat) := match x, y with | 0, 0 => 1 | x+1, 0 => f x 0 | 0, y+1 => f 0 y | x+1, y+1 => f x y termination_by f x y => (x, y) example (x y : Nat) : f x y > 0 := by induction x, y <;> simp [f, *]
Add support for casesOn applications to structural and well-founded recursion modules. This feature is useful when writing definitions using tactics. Example:
inductive Foo where | a | b | c | pair: Foo × Foo → Foo def Foo.deq (a b : Foo) : Decidable (a = b) := by cases a <;> cases b any_goals apply isFalse Foo.noConfusion any_goals apply isTrue rfl case pair a b => let (a₁, a₂) := a let (b₁, b₂) := b exact match deq a₁ b₁, deq a₂ b₂ with | isTrue h₁, isTrue h₂ => isTrue (by rw [h₁,h₂]) | isFalse h₁, _ => isFalse (fun h => by cases h; cases (h₁ rfl)) | _, isFalse h₂ => isFalse (fun h => by cases h; cases (h₂ rfl))
Option is again a monad. The auxiliary type OptionM has been removed. See Zulip thread.
Improve split tactic. It used to fail on match expressions of the form match h : e with ... where e is not a free variable. The failure used to occur during generalization.
New encoding for match-expressions that use the h : notation for discriminants. The information is not lost during delaboration, and it is the foundation for a better split tactic. at delaboration time. Example:
#print Nat.decEq /- protected def Nat.decEq : (n m : Nat) → Decidable (n = m) := fun n m => match h : Nat.beq n m with | true => isTrue (_ : n = m) | false => isFalse (_ : ¬n = m) -/
exists tactic is now takes a comma separated list of terms.
Add dsimp and dsimp! tactics. They guarantee the result term is definitionally equal, and only apply rfl-theorems.
Fix binder information for match patterns that use definitions tagged with [matchPattern] (e.g., Nat.add). We now have proper binder information for the variable y in the following example.
def f (x : Nat) : Nat := match x with | 0 => 1 | y + 1 => y
(Fix) the default value for structure fields may now depend on the structure parameters. Example:
structure Something (i: Nat) where n1: Nat := 1 n2: Nat := 1 + i def s : Something 10 := {} example : s.n2 = 11 := rfl
Apply rfl theorems at the dsimp auxiliary method used by simp. dsimp can be used anywhere in an expression because it preserves definitional equality.
Refine auto bound implicit feature. It does not consider anymore unbound variables that have the same name of a declaration being defined. Example:
def f : f → Bool := -- Error at second `f` fun _ => true inductive Foo : List Foo → Type -- Error at second `Foo` | x : Foo []
Before this refinement, the declarations above would be accepted and the second f and Foo would be treated as auto implicit variables. That is, f : {f : Sort u} → f → Bool, and Foo : {Foo : Type u} → List Foo → Type.
Fix syntax highlighting for recursive declarations. Example
inductive List (α : Type u) where | nil : List α -- `List` is not highlighted as a variable anymore | cons (head : α) (tail : List α) : List α def List.map (f : α → β) : List α → List β | [] => [] | a::as => f a :: map f as -- `map` is not highlighted as a variable anymore
Add autoUnfold option to Lean.Meta.Simp.Config, and the following macros
simp! for simp (config := { autoUnfold := true })simp_arith! for simp (config := { autoUnfold := true, arith := true })simp_all! for simp_all (config := { autoUnfold := true })simp_all_arith! for simp_all (config := { autoUnfold := true, arith := true })When the autoUnfold is set to true, simp tries to unfold the following kinds of definition
match-expression. This kind of definition is only unfolded if the match can be reduced. Example:def append (as bs : List α) : List α := match as with | [] => bs | a :: as => a :: append as bs theorem append_nil (as : List α) : append as [] = as := by induction as <;> simp_all! theorem append_assoc (as bs cs : List α) : append (append as bs) cs = append as (append bs cs) := by induction as <;> simp_all!
Add save tactic for creating checkpoints more conveniently. Example:
example : <some-proposition> := by tac_1 tac_2 save tac_3 ...
is equivalent to
example : <some-proposition> := by checkpoint tac_1 tac_2 tac_3 ...
Remove support for {} annotation from inductive datatype constructors. This annotation was barely used, and we can control the binder information for parameter bindings using the new inductive family indices to parameter promotion. Example: the following declaration using {}
inductive LE' (n : Nat) : Nat → Prop where | refl {} : LE' n n -- Want `n` to be explicit | succ : LE' n m → LE' n (m+1)
can now be written as
inductive LE' : Nat → Nat → Prop where | refl (n : Nat) : LE' n n | succ : LE' n m → LE' n (m+1)
In both cases, the inductive family has one parameter and one index. Recall that the actual number of parameters can be retrieved using the command #print.
Remove support for {} annotation in the structure command.
Several improvements to LSP server. Examples: “jump to definition” in mutually recursive sections, fixed incorrect hover information in “match”-expression patterns, “jump to definition” for pattern variables, fixed auto-completion in function headers, etc.
In macro ... xs:p* ... and similar macro bindings of combinators, xs now has the correct type Array Syntax
Identifiers in syntax patterns now ignore macro scopes during matching.
Improve binder names for constructor auto implicit parameters. Example, given the inductive datatype
inductive Member : α → List α → Type u | head : Member a (a::as) | tail : Member a bs → Member a (b::bs)
before:
#check @Member.head -- @Member.head : {x : Type u_1} → {a : x} → {as : List x} → Member a (a :: as)
now:
#check @Member.head -- @Member.head : {α : Type u_1} → {a : α} → {as : List α} → Member a (a :: as)
Improve error message when constructor parameter universe level is too big.
Add support for for h : i in [start:stop] do .. where h : i ∈ [start:stop]. This feature is useful for proving termination of functions such as:
inductive Expr where | app (f : String) (args : Array Expr) def Expr.size (e : Expr) : Nat := Id.run do match e with | app f args => let mut sz := 1 for h : i in [: args.size] do -- h.upper : i < args.size sz := sz + size (args.get ⟨i, h.upper⟩) return sz
Add tactic case'. It is similar to case, but does not admit the goal on failure. For example, the new tactic is useful when writing tactic scripts where we need to use case' at first | ... | ..., and we want to take the next alternative when case' fails.
Add tactic macro
macro "stop" s:tacticSeq : tactic => `(repeat sorry)
See discussion on Zulip.
When displaying goals, we do not display inaccessible proposition names if they do not have forward dependencies. We still display their types. For example, the goal
case node.inl.node β : Type u_1 b : BinTree β k : Nat v : β left : Tree β key : Nat value : β right : Tree β ihl : BST left → Tree.find? (Tree.insert left k v) k = some v ihr : BST right → Tree.find? (Tree.insert right k v) k = some v h✝ : k < key a✝³ : BST left a✝² : ForallTree (fun k v => k < key) left a✝¹ : BST right a✝ : ForallTree (fun k v => key < k) right ⊢ BST left
is now displayed as
case node.inl.node β : Type u_1 b : BinTree β k : Nat v : β left : Tree β key : Nat value : β right : Tree β ihl : BST left → Tree.find? (Tree.insert left k v) k = some v ihr : BST right → Tree.find? (Tree.insert right k v) k = some v : k < key : BST left : ForallTree (fun k v => k < key) left : BST right : ForallTree (fun k v => key < k) right ⊢ BST left
The hypothesis name is now optional in the by_cases tactic.
Fix inconsistency between syntax and kind names. The node kinds numLit, charLit, nameLit, strLit, and scientificLit are now called num, char, name, str, and scientific respectively. Example: we now write
macro_rules | `($n:num) => `("hello")
instead of
macro_rules | `($n:numLit) => `("hello")
(Experimental) New checkpoint <tactic-seq> tactic for big interactive proofs.
Rename tactic nativeDecide => native_decide.
Antiquotations are now accepted in any syntax. The incQuotDepth syntax parser is therefore obsolete and has been removed.
Renamed tactic nativeDecide => native_decide.
“Cleanup” local context before elaborating a match alternative right-hand-side. Examples:
example (x : Nat) : Nat := match g x with | (a, b) => _ -- Local context does not contain the auxiliary `_discr := g x` anymore example (x : Nat × Nat) (h : x.1 > 0) : f x > 0 := by match x with | (a, b) => _ -- Local context does not contain the `h✝ : x.fst > 0` anymore
Improve let-pattern (and have-pattern) macro expansion. In the following example,
example (x : Nat × Nat) : f x > 0 := by let (a, b) := x done
The resulting goal is now ... |- f (a, b) > 0 instead of ... |- f x > 0.
Add cross-compiled aarch64 Linux and aarch64 macOS releases.
Add tutorial-like examples to our documentation, rendered using LeanInk+Alectryon.
simp now takes user-defined simp-attributes. You can define a new simp attribute by creating a file (e.g., MySimp.lean) containing
import Lean open Lean.Meta initialize my_ext : SimpExtension ← registerSimpAttr `my_simp "my own simp attribute"
If you don't need to access my_ext, you can also use the macro
import Lean register_simp_attr my_simp "my own simp attribute"
Recall that the new simp attribute is not active in the Lean file where it was defined. Here is a small example using the new feature.
import MySimp def f (x : Nat) := x + 2 def g (x : Nat) := x + 1 @[my_simp] theorem f_eq : f x = x + 2 := rfl @[my_simp] theorem g_eq : g x = x + 1 := rfl example : f x + g x = 2*x + 3 := by simp_arith [my_simp]
Extend match syntax: multiple left-hand-sides in a single alternative. Example:
def fib : Nat → Nat | 0 | 1 => 1 | n+2 => fib n + fib (n+1)
This feature was discussed at issue 371. It was implemented as a macro expansion. Thus, the following is accepted.
inductive StrOrNum where | S (s : String) | I (i : Int) def StrOrNum.asString (x : StrOrNum) := match x with | I a | S a => toString a
Improve #eval command. Now, when it fails to synthesize a Lean.MetaEval instance for the result type, it reduces the type and tries again. The following example now works without additional annotations
def Foo := List Nat def test (x : Nat) : Foo := [x, x+1, x+2] #eval test 4
rw tactic can now apply auto-generated equation theorems for a given definition. Example:
example (a : Nat) (h : n = 1) : [a].length = n := by rw [List.length] trace_state -- .. |- [].length + 1 = n rw [List.length] trace_state -- .. |- 0 + 1 = n rw [h]
Extend dot-notation x.field for arrow types. If type of x is an arrow, we look up for Function.field. For example, given f : Nat → Nat and g : Nat → Nat, f.comp g is now notation for Function.comp f g.
The new .<identifier> notation is now also accepted where a function type is expected.
example (xs : List Nat) : List Nat := .map .succ xs example (xs : List α) : Std.RBTree α ord := xs.foldl .insert ∅
Support notation let <pattern> := <expr> | <else-case> in do blocks.
Remove support for “auto” pure. In the Zulip thread, the consensus seemed to be that “auto” pure is more confusing than it's worth.
Remove restriction in congr theorems that all function arguments on the left-hand-side must be free variables. For example, the following theorem is now a valid congr theorem.
@[congr] theorem dep_congr [DecidableEq ι] {p : ι → Set α} [∀ i, Inhabited (p i)] : ∀ {i j} (h : i = j) (x : p i) (y : α) (hx : x = y), Pi.single (f := (p ·)) i x = Pi.single (f := (p ·)) j ⟨y, hx ▸ h ▸ x.2⟩ :=
Improve elaboration postponement heuristic when expected type is a metavariable. Lean now reduces the expected type before performing the test.
Remove deprecated leanpkg in favor of Lake now bundled with Lean.
Various improvements to go-to-definition & find-all-references accuracy.
Auto generated congruence lemmas with support for casts on proofs and Decidable instances (see wishlist).
Rename option autoBoundImplicitLocal => autoImplicit.
Relax auto-implicit restrictions. The command set_option relaxedAutoImplicit false disables the relaxations.
contradiction tactic now closes the goal if there is a False.elim application in the target.
Renamed tatic byCases => by_cases (motivation: enforcing naming convention).
Local instances occurring in patterns are now considered by the type class resolution procedure. Example:
def concat : List ((α : Type) × ToString α × α) → String | [] => "" | ⟨_, _, a⟩ :: as => toString a ++ concat as
Notation for providing the motive for match expressions has changed. before:
match x, rfl : (y : Nat) → x = y → Nat with | 0, h => ... | x+1, h => ...
now:
match (motive := (y : Nat) → x = y → Nat) x, rfl with | 0, h => ... | x+1, h => ...
With this change, the notation for giving names to equality proofs in match-expressions is not whitespace sensitive anymore. That is, we can now write
match h : sort.swap a b with | (r₁, r₂) => ... -- `h : sort.swap a b = (r₁, r₂)`
(generalizing := true) is the default behavior for match expressions even if the expected type is not a proposition. In the following example, we used to have to include (generalizing := true) manually.
inductive Fam : Type → Type 1 where | any : Fam α | nat : Nat → Fam Nat example (a : α) (x : Fam α) : α := match x with | Fam.any => a | Fam.nat n => n
We now use PSum (instead of Sum) when compiling mutually recursive definitions using well-founded recursion.
Better support for parametric well-founded relations. See issue #1017. This change affects the low-level termination_by' hint because the fixed prefix of the function parameters in not “packed” anymore when constructing the well-founded relation type. For example, in the following definition, as is part of the fixed prefix, and is not packed anymore. In previous versions, the termination_by' term would be written as measure fun ⟨as, i, _⟩ => as.size - i
def sum (as : Array Nat) (i : Nat) (s : Nat) : Nat := if h : i < as.size then sum as (i+1) (s + as.get ⟨i, h⟩) else s termination_by' measure fun ⟨i, _⟩ => as.size - i
Add while <cond> do <do-block>, repeat <do-block>, and repeat <do-block> until <cond> macros for do-block. These macros are based on partial definitions, and consequently are useful only for writing programs we don't want to prove anything about.
Add arith option to Simp.Config, the macro simp_arith expands to simp (config := { arith := true }). Only Nat and linear arithmetic is currently supported. Example:
example : 0 < 1 + x ∧ x + y + 2 ≥ y + 1 := by simp_arith
Add fail <string>? tactic that always fail.
Add support for acyclicity at dependent elimination. See issue #1022.
Add trace <string> tactic for debugging purposes.
Add nontrivial SizeOf instance for types Unit → α, and add support for them in the auto-generated SizeOf instances for user-defined inductive types. For example, given the inductive datatype
inductive LazyList (α : Type u) where | nil : LazyList α | cons (hd : α) (tl : LazyList α) : LazyList α | delayed (t : Thunk (LazyList α)) : LazyList α
we now have sizeOf (LazyList.delayed t) = 1 + sizeOf t instead of sizeOf (LazyList.delayed t) = 2.
Add support for guessing (very) simple well-founded relations when proving termination. For example, the following function does not require a termination_by annotation anymore.
def Array.insertAtAux (i : Nat) (as : Array α) (j : Nat) : Array α := if h : i < j then let as := as.swap! (j-1) j; insertAtAux i as (j-1) else as
Add support for for h : x in xs do ... notation where h : x ∈ xs. This is mainly useful for showing termination.
Auto implicit behavior changed for inductive families. An auto implicit argument occurring in inductive family index is also treated as an index (IF it is not fixed, see next item). For example
inductive HasType : Index n → Vector Ty n → Ty → Type where
is now interpreted as
inductive HasType : {n : Nat} → Index n → Vector Ty n → Ty → Type where
To make the previous feature more convenient to use, we promote a fixed prefix of inductive family indices to parameters. For example, the following declaration is now accepted by Lean
inductive Lst : Type u → Type u | nil : Lst α | cons : α → Lst α → Lst α
and α in Lst α is a parameter. The actual number of parameters can be inspected using the command #print Lst. This feature also makes sure we still accept the declaration
inductive Sublist : List α → List α → Prop | slnil : Sublist [] [] | cons l₁ l₂ a : Sublist l₁ l₂ → Sublist l₁ (a :: l₂) | cons2 l₁ l₂ a : Sublist l₁ l₂ → Sublist (a :: l₁) (a :: l₂)
Added auto implicit “chaining”. Unassigned metavariables occurring in the auto implicit types now become new auto implicit locals. Consider the following example:
inductive HasType : Fin n → Vector Ty n → Ty → Type where | stop : HasType 0 (ty :: ctx) ty | pop : HasType k ctx ty → HasType k.succ (u :: ctx) ty
ctx is an auto implicit local in the two constructors, and it has type ctx : Vector Ty ?m. Without auto implicit “chaining”, the metavariable ?m will remain unassigned. The new feature creates yet another implicit local n : Nat and assigns n to ?m. So, the declaration above is shorthand for
inductive HasType : {n : Nat} → Fin n → Vector Ty n → Ty → Type where | stop : {ty : Ty} → {n : Nat} → {ctx : Vector Ty n} → HasType 0 (ty :: ctx) ty | pop : {n : Nat} → {k : Fin n} → {ctx : Vector Ty n} → {ty : Ty} → HasType k ctx ty → HasType k.succ (u :: ctx) ty
Eliminate auxiliary type annotations (e.g, autoParam and optParam) from recursor minor premises and projection declarations. Consider the following example
structure A := x : Nat h : x = 1 := by trivial example (a : A) : a.x = 1 := by have aux := a.h -- `aux` has now type `a.x = 1` instead of `autoParam (a.x = 1) auto✝` exact aux example (a : A) : a.x = 1 := by cases a with | mk x h => -- `h` has now type `x = 1` instead of `autoParam (x = 1) auto✝` assumption
We now accept overloaded notation in patterns, but we require the set of pattern variables in each alternative to be the same. Example:
inductive Vector (α : Type u) : Nat → Type u | nil : Vector α 0 | cons : α → Vector α n → Vector α (n+1) infix:67 " :: " => Vector.cons -- Overloading the `::` notation def head1 (x : List α) (h : x ≠ []) : α := match x with | a :: as => a -- `::` is `List.cons` here def head2 (x : Vector α (n+1)) : α := match x with | a :: as => a -- `::` is `Vector.cons` here
New notation .<identifier> based on Swift. The namespace is inferred from the expected type. See issue #944. Examples:
def f (x : Nat) : Except String Nat := if x > 0 then .ok x else .error "x is zero" namespace Lean.Elab open Lsp def identOf : Info → Option (RefIdent × Bool) | .ofTermInfo ti => match ti.expr with | .const n .. => some (.const n, ti.isBinder) | .fvar id .. => some (.fvar id, ti.isBinder) | _ => none | .ofFieldInfo fi => some (.const fi.projName, false) | _ => none def isImplicit (bi : BinderInfo) : Bool := bi matches .implicit end Lean.Elab