This PR moves the elaboration of structure/class Verso docstrings until
after the fact that it's a class is registered, so code samples in the
docstring can use it as a class. Redundant addition of structure and
constructor docstrings are also removed, because they're already handled
in MutualInductive.lean.
Closes#11651
This PR uses an `Array` instead of a `List` to store the clauses in
`Std.CNF`. This reduces the memory footprint and pressure on the
allocator, leading to noticeable performance changes with gigantic CNFs.
This PR treats the first character of the first line of a docstring as
being in the leftmost column, even if it physically is not. This allows
left-column items like headers to be used even after spaces. It also
detects the indentation of the entire docstring, using it as the
zero-point for indentation sensitive syntax such as headers.
Closes#12067.
This PR fixes poor error reporting from Verso docstrings. Before, if the
Verso parser didn't consume the whole docstring, then Lean would try to
parse the closing -/ and fail; this would lead to backtracking and an
assumption that the docstring must be non-Verso, with only the non-Verso
commands like #guard_msgs as possibilities. Now, the input is always
consumed.
Closes#12118.
This PR implements better support for unfolding class fields marked as
`reducible`. For example, we want to mark fields that are types such as
```lean
MonadControlT.stM : Type u -> Type u
```
The motivation is similar to our heuristic that type definitions should
be abbreviations.
Now, suppose we want to unfold `stM m (ExceptT ε m) α` using the
`.reducible` transparency setting, we want the result to be `stM m m
(MonadControl.stM m (ExceptT ε m) α)` instead of
`(instMonadControlTOfMonadControl m m (ExceptT ε m)).1 α`. The latter
would defeat the intent of marking the field as reducible, since the
instance `instMonadControlTOfMonadControl` is `[instance_reducible]` and
the resulting term would be stuck when using `.reducible` transparency
mode.
**Remark**: This feature introduces a few breakages in core and Mathlib.
So, it is disabled for now in this PR. To enable, we must use
`set_option backward.whnf.reducibleClassField true`
This PR adds two benchmarks (sieve of Eratosthenes, removing duplicates
from the list) and one test (a function with sublinear complexity
defined via well-founded recursion evaluated on large naturals with up
to `60` digits).
The tests have been suggested by @b-mehta.
This PR changes the semantics of `inline` annotations in the compiler.
The behavior of the original `@[inline]` attribute remains the same but
the function `inline` now comes with a restriction that it can only use
declarations that are local to the current module. This comes as a
preparation to pulling the compiler out into a separate process.
Closes: #12334
This PR changes the handling of over-applied cases expressions in
`ToLCNF` to avoid generating function declarations that are called
immediately. For example, `ToLCNF` previously produced this:
```lean-4
set_option trace.Compiler.init true
/--
trace: [Compiler.init] size: 4
def test x y : Bool :=
fun _y.1 _y.2 : Bool :=
cases x : Bool
| PUnit.unit =>
fun _f.3 a : Bool :=
return a;
let _x.4 := _f.3 _y.2;
return _x.4;
let _x.5 := _y.1 y;
return _x.5
-/
#guard_msgs in
def test (x : Unit) (y : Bool) : Bool :=
x.casesOn (fun a => a) y
```
which is now simplified to
```lean-4
set_option trace.Compiler.init true
/--
trace: [Compiler.init] size: 3
def test x y : Bool :=
cases x : Bool
| PUnit.unit =>
let a := y;
return a
-/
#guard_msgs in
def test (x : Unit) (y : Bool) : Bool :=
x.casesOn (fun a => a) y
```
This is especially relevant for #8309 because there `dite` is defined as
an over-applied `Bool.casesOn`.
This PR adds the option `doc.verso.module`. If set, it controls whether
module docstrings use Verso syntax. If not set, it defaults to the value
of the `doc.verso` option.
Closes#12070.
This PR fixes an issue on new NeoVim versions that would cause the
language server to display an error when using certain code actions.
(For some reason, NeoVim recently decided to diverge from VS Code in
terms of when it emits code action resolution requests, which means that
not setting the `data?` field won't preclude NeoVim from emitting a
request anymore, which in turn means that the server can't resolve the
code action.)
This PR implements preparatory work for #12179. It implements a new
feature in `isDefEq` to ensure it does not increase the transparency
level to `.default` when checking definitionally equality of implicit
arguments. This transparency level bump was introduced in Lean 3, but it
is not a performance issue and is affecting Mathlib. This PR adds the
new feature, but it is disabled by default.
This PR adds a default `Inhabited` instance to `Theorem` type.
The need to do so came up in #12296 , as `Theorem` is one of the entries
of the structure which is the key entry of `SimpleScopedEnvExtension`.
This PR fixes the PR release workflow which was failing due to the
deprecated `MATHLIB4_COMMENT_BOT` PAT being invalid.
**Error:** `jq: error (at <stdin>:4): Cannot index string with string
"name"` when fetching labels - the API returned an error response
instead of labels array because the PAT credentials were bad.
**Changes:**
- Generate a token from the `mathlib-lean-pr-testing` GitHub App (ID:
2785182) for posting comments to Lean PRs about mathlib compatibility
- Use `GITHUB_TOKEN` for read-only label fetching (no special identity
needed)
- Update the bot username check from `leanprover-community-bot` to
`mathlib-lean-pr-testing[bot]`
**Secrets added:** `MATHLIB_LEAN_PR_TESTING_APP_ID` and
`MATHLIB_LEAN_PR_TESTING_PRIVATE_KEY` have been added to the repository.
Fixes the CI failure at
https://github.com/leanprover/lean4/actions/runs/21705647318/job/62595966115🤖 Prepared with Claude Code
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
This PR adds `String.Slice.Subslice`, which is an unbundled version of
`String.Slice`.
This type is of interest because it is the correct type for string
searching and splitting operations to land in.
This PR just adds the type with minimal API. Additional API and
subsequent refactoring of the searching and splitting API is left for
future PRs.
This PR reverses the relationship between the `ForwardPattern` and
`ToForwardSearcher` classes.
Previously, it was possible to derive `ForwardPattern` (i.e.,
`dropPrefix?`) from `ToForwardSearcher` (i.e., get an iterator of
`SearchStep (s)`). Now, we give the default instance in the other
direction: it is now possible to derive `ToForwardSearcher` from
`ForwardPattern`. Since it is usually much easier to provide
`ForwardPattern` than `ToForwardSearcher`, this means more shared code,
which pays off double since we will give a correctness proof for the
default implementation in an upcoming PR.
This PR also adds some string lemmas.
This PR adds theorems that directly state that div and mod form an
injective pair: if `a / n = b / n` and `a % n = b % n` then `a = b`.
These complement existing div/mod lemmas and are useful for extension
arguments.
Upstreaming from
https://github.com/leanprover-community/mathlib4/pull/34201🤖 Prepared with Claude Code
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
This PR fixes the PR release workflow which is failing to create
`lean-pr-testing` branches due to the deprecated `MATHLIB4_BOT` PAT (see
https://github.com/leanprover/lean4/actions/runs/21698772126/job/62574742680).
The fix uses `actions/create-github-app-token@v2` to generate a token
from the `mathlib-nightly-testing` GitHub App (ID: 2784211) instead,
which has write access to both `leanprover-community/batteries` and
`leanprover-community/mathlib4-nightly-testing`.
Requires adding `MATHLIB_NIGHTLY_TESTING_APP_ID` and
`MATHLIB_NIGHTLY_TESTING_PRIVATE_KEY` secrets to leanprover/lean4
(done).
🤖 Prepared with Claude Code
This PR ensures `simp` does not "simplify" instances by default. The old
behavior can be retrieved by using `simp +instances`. This PR is similar
to #12195, but for `dsimp`.
The backward compatibility flag for `dsimp` also deactivates this new
feature.
```
set_option backward.dsimp.instances true
```
Applying `simp` (and `dsimp`) to instances creates non-standard
instances, and this creates all sorts of problems in Mathlib.
---------
Co-authored-by: Henrik Böving <hargonix@gmail.com>
Co-authored-by: Sebastian Graf <sgraf1337@gmail.com>
Co-authored-by: Kim Morrison <kim@tqft.net>
This PR leverages the fact that expressions are type correct in `grind`
and the conclusion of extensionality theorems is of the form `?a = ?b`.
This PR is relevant for #12179 because it enables us to use a weaker
`isDefEq` that does not bump the transparency level when processing
implicit arguments.
