the internal constructions for structural and well-founded recursion
use plenty of `PProd` and `MProd`, and reading these, deeply
nested and in prefix notation, is unnecessarily troublesome.
Therefore this introduces notations
```
a ×ₚ b -- PProd a b
a ×ₘ b -- MProd a b
()ₚ -- PUnit.unit
(x,y,z)ₚ -- PProd.mk x (PProd.mk y z)
(x,y,z)ₘ -- MProd.mk x (MProd.mk y z)
```
(This is part 1, the rest will follow in #4730 after a stage0 update.)
Generalizes #3556 to not suppressing errors in tactic steps either when
the parse error is in a later step, as otherwise changes to the end of a
proof would affect (correctness or effectiveness of) incrementality of
preceding steps.
Fixes#4623, in combination with #4643
This implements the `termination_by structural` syntax proposed in
#3909.
I went with `termination_by structural` over, say,
`termination_by (config := {method := .structural})` mainly because it
was
easier to get going (otherwise I’d have to look into how to define
recursive
parsers, as `Parser.config` depends on `term` and `termination_by` is
part of
term. But also because I find it more ergonomic and aesthetic as a user.
But syntax can still change.
The `termination_by?` syntax will no longer force well-founded
recursion,
and instead the inferred `termination_by structurally` annotation will
be shown
if structural termination is possible.
While I was it, this fixes#4546 the easy way (log errors about but
otherwise
ignore incomplete `termination_by` sets for mutual recursion). Maybe we
get
multiple replacements (#4551), but even then this this good behavior.
Involves a bit of shuffling around `TerimationHints` (now validated for
a
clique already by `PreDefinition.main`) and `TerminationArguments` (now
lifted
out of the `WF` namespace, and a bit simplified).
Fixes#3909
---------
Co-authored-by: Richard Kiss <him@richardkiss.com>
This appears to have been a semantic merge conflict between #3940 and
#4129. The effect on the language server is that if two edits are
sufficiently close in time to create an interrupt, some elaboration
steps like `simp` may accidentally catch the exception when it is
triggered during their execution, which makes incrementality assume that
elaboration of the body was successful, which can lead to incorrect
reuse, presenting the interrupted state to the user with symptoms such
as "uses sorry" without accompanying errors and incorrect lints.
When the type of an `example` is a proposition,
we should elaborate on them as we elaborate on theorems.
This is particularly important for examples that are often
used in educational material.
Recall that when elaborating theorem headers, we convert unassigned
universe metavariables into universe parameters. The motivation is
that the proof of a theorem should not influence its statement.
However, before this commit, this was not the case for examples when
their type was a proposition.
This discrepancy often confused users.
Additionally, we considered extending the above behavior to definitions
when
1- When their type is a proposition. However, it still caused disruption
in Mathlib.
2- When their type is provided. That is, we would keep the current
behavior only if `: <type>` was omitted. This would make the elaborator
for `def` much closer to the one for `theorem`, but it proved to be too
restrictive.
For example, the following instance in `Core.lean` would fail:
```
instance {α : Sort u} [Setoid α] : HasEquiv α :=
⟨Setoid.r⟩
```
and we would have to write instead:
```
instance {α : Sort u} [Setoid α] : HasEquiv.{u, 0} α :=
⟨Setoid.r⟩
```
There are other failures like this in the core, and we assume many more
in Mathlib.
closes#4398closes#4482 Remark: PR #4482 implements option 1 above. We may consider
it again in the future.
as #4527 describes there is inconsistency between `by`, `case` and
`next` on the one hand who, if the goal isn’t closed, put squiggly
underlines on the first line, and `.`, which so far only squiggled the
dot (which is a very short symbol!)
With this change the same mechanism as used by `case`, namely
`withCaseRef`, is also used for `.`.
There is an argument for the status quo: The `.` tactic is more commonly
used
with further tactics on the same line, and thus there is now a higher
risk that
the user might think that the first tactic is broken. But
* the same argument does apply to `by` and `case` where there was an
intentional
choice to do it this way
* consistency and
* a squiggly line just under the short `.` is easy to miss, so it is
actually
better to underlining more here (at least until we have a better way to
indicate incomplete proofs, which I have hopes for)
Fixes#4527, at least most of it.
This is the groundwork for a tactic index in generated documentation, as
there was in Lean 3. There are a few challenges to getting this to work
well in Lean 4:
* There's no natural notion of *tactic identity* - a tactic may be
specified by multiple syntax rules (e.g. the pattern-matching version of
`intro` is specified apart from the default version, but both are the
same from a user perspective)
* There's no natural notion of *tactic name* - here, we take the
pragmatic choice of using the first keyword atom in the tactic's syntax
specification, but this may need to be overridable someday.
* Tactics are extensible, but we don't want to allow arbitrary imports
to clobber existing tactic docstrings, which could become unpredictable
in practice.
For tactic identity, this PR introduces the notion of a *tactic
alternative*, which is a `syntax` specification that is really "the same
as" an existing tactic, but needs to be separate for technical reasons.
This provides a notion of tactic identity, which we can use as the basis
of a tactic index in generated documentation. Alternative forms of
tactics are specified using a new `@[tactic_alt IDENT]` attribute,
applied to the new tactic syntax. It is an error to declare a tactic
syntax rule to be an alternative of another one that is itself an
alternative. Documentation hovers now take alternatives into account,
and display the docs for the canonical name.
*Tactic tags*, created with the `register_tactic_tag` command, specify
tags that may be applied to tactics. This is intended to be used by
doc-gen and Verso. Tags may be applied using the `@[tactic_tag TAG1 TAG2
...]` attribute on a canonical tactic parser, which may be used in any
module to facilitate downstream projects introducing tags that apply to
pre-existing tactics. Tags may not be removed, but it's fine to
redundantly add them. The collection of tags, and the tactics to which
they're applied, can be seen using the `#print tactic tags` command.
*Extension documentation* provides a structured way to document
extensions to tactics. The resulting documentation is gathered into a
bulleted list at the bottom of the tactic's docstring. Extensions are
added using the `tactic_extension TAC` command. This can be used when
adding new interpretations of a tactic via `macro_rules`, when extending
some table or search index used by the tactic, or in any other way. It
is a command to facilitate its flexible use with various extension
mechanisms.
`Nat.succ_eq_add_one` and `Nat.pred_eq_sub_one` are now simp lemmas. For
theorems about `Nat.succ` or `Nat.pred` without corresponding theorem
for `+ 1` or `- 1`, this adds the corresponding theorem.
A pending tactic mvar managed to escape into an unexpected context in
specific circumstances.
```lean
example : True := by
· rw [show 0 = 0 by rfl]
```
* Term elaboration of the `show` creates a pending mvar for the `by rfl`
proof
* `rw` fails with an exception because the pattern does not occur in the
target
* `cdot` catches the exception and admits the goal
* `Term.runTactic` [synthesizes all pending mvars from the tactic's
execution](5f9dedfe5e/src/Lean/Elab/SyntheticMVars.lean (L350)),
including the `by rfl` proof. But this would not have happened without
`cdot` as the exception would have skipped that invocation!
* Now incrementality is confused because the nested `by rfl` proof is
unexpectedly run in the same context as the top-level proof, writing to
the wrong promise, and the error message is lost
Solution: disable incrementality for these pending mvars
As [reported on
Zulip](https://leanprover.zulipchat.com/#narrow/stream/113488-general/topic/maybe.20a.20cache.20bug.3F).
We expected that for sound reuse of elaboration results, it is
sufficient to compare the old and new syntax tree's structure and atoms
including position info, but not the whitespace in between them.
However, we have at least one request handler, the goal view, that
inspects the whitespace after a tactic and thus could return incorrect
results on reuse. For now we implement the straightforward fix of
checking the whitespace as well. Alternatives like updating the
whitespace stored in the reused info tree are tbd.
This has the slight disadvantage that adding whitespace at the end of a
tactic will re-execute it (or the entire body, but not the header, if
the body is not a tactic block), but only up to typing the first
character of the next tactic or command.
this is an amendment to #4177, after @kmill pointed out an issue:
Users might expect that within a tactic combinator like `first`, `simp
[h]` fails if `h` does not exist. Therefore the behavior introduced in
PR #4177, which is really most useful in mormal interactive use of
`skip`, is restricted to when `recover := true`.
The `save` happened in a slightly different context from the restore,
which a refinement of the `saveOrRestoreFull` signature now makes
impossible.
Fixes#4328
this fixes a usability paper cut that just annoyed me. When editing a
larger simp proof, I usually want to see the goal state after the simp,
and this is what I see while the `simp` command is complete. But then,
when I start typing, and necessarily type incomplete lemma names, that
error makes `simp` do nothing again and I see the original goal state.
In fact, if a prefix of the simp theorem name I am typing is a valid
identifier, it jumps even more around.
With this PR, using `logException`, I still get the red squiggly lines
for the unknown identifer, but `simp` just ignores that argument and
still shows me the final goal. Much nicer.
I also demoted the message for `[-foo]` when `foo` isn’t `simp` to a
warning and gave it the correct `ref`.
See it in action here: (in the middle, when you suddenly see the
terminal,
I am switching lean versions.)
https://github.com/leanprover/lean4/assets/148037/8cb3c563-1354-4c2d-bcee-26dfa1005ae0
Without this, it would not easy but perhaps be feasible to break
incrementality when editing command prefixes such as `set_option ... in
theorem` or also `theorem namesp.name ...` (which is a macro),
especially if at some later point we support incrementality in input
shifted by an edit. Explicit, sound support for these common cases will
be brought back soon.
Extends Lean's incremental reporting and reuse between commands into
various steps inside declarations:
* headers and bodies of each (mutual) definition/theorem
* `theorem ... := by` for each contained tactic step, including
recursively inside supported combinators currently consisting of
* `·` (cdot), `case`, `next`
* `induction`, `cases`
* macros such as `next` unfolding to the above
*Incremental reuse* means not recomputing any such steps if they are not
affected by a document change. *Incremental reporting* includes the
parts seen in the recording above: the progress bar and messages. Other
language server features such as hover etc. are *not yet* supported
incrementally, i.e. they are shown only when the declaration has been
fully processed as before.
---------
Co-authored-by: Scott Morrison <scott.morrison@gmail.com>
this fixes#4078. It is an alternative fix to the one in #4137,
suggested
by @kmill.
Incidentially, it makes the unused variable linter better. My theory is
that
if we don’t reset the info when backtracking, the binder shows up more
than
once in the info tree, and then it is considered “used”, although there
are
just multiple binders.
This coercion caused difficult-to-diagnose bugs sometimes. Because there
are some situations where converting a string to a name should be done
by parsing the string, and others where it should not, an explicit
choice seems better here.
---------
Co-authored-by: Mac Malone <tydeu@hatpress.net>
This is a rewrite of the `UnusedVariables` lint to inline and simplify
many of the dependent functions to try to improve the performance of
this lint, which quite often shows up in perf reports.
* The mvar assignment scanning is one of the most expensive parts of the
process, so we do two things to improve this:
* Lazily perform the scan only if we need it
* Use an object-pointer hashmap to ensure that we don't have quadratic
behavior when there are many mvar assignments with slight differences.
* The dependency on `Lean.Server` is removed, meaning we don't need to
do the LSP conversion stuff anymore. The main logic of reference finding
is inlined.
* We take `fvarAliases` into account, and union together fvars which are
aliases of a base fvar. (It would be great if we had `UnionFind` here.)
More docs will be added once we confirm an actual perf improvement.
---------
Co-authored-by: Sebastian Ullrich <sebasti@nullri.ch>
this makes `termination_by?` even slicker.
The heuristics is agressive in the non-mutual case (will omit `sizeOf`
if the argument is non-dependent and the `WellFoundedRelation` relation
is via `sizeOfWFRel`.
In the mutual case we'd also have to check the arguments, as they line
up in the termination argument, have the same types. I did not bother at
this point; in the mutual case we omit `sizeOf` only if the argument
type is `Nat`.
As a drive-by fix, `termination_by?` now also works on functions that
have only one plausible measure.
Replaces `@[eliminator]` with two attributes `@[induction_eliminator]`
and `@[cases_eliminator]` for defining custom eliminators for the
`induction` and `cases` tactics, respectively.
Adds `Nat.recAux` and `Nat.casesAuxOn`, which are eliminators that are
defeq to `Nat.rec` and `Nat.casesOn`, but these use `0` and `n + 1`
rather than `Nat.zero` and `Nat.succ n`.
For example, using `induction` to prove that the factorial function is
positive now has the following goal states (thanks also to #3616 for the
goal state after unfolding).
```lean
example : 0 < fact x := by
induction x with
| zero => decide
| succ x ih =>
/-
x : Nat
ih : 0 < fact x
⊢ 0 < fact (x + 1)
-/
unfold fact
/-
...
⊢ 0 < (x + 1) * fact x
-/
simpa using ih
```
Thanks to @adamtopaz for initial work on splitting the `@[eliminator]`
attribute.
The `delabConstWithSignature` delaborator is responsible for pretty
printing constants with a declaration-like signature, with binders, a
colon, and a type. This is used by the `#check` command when it is given
just an identifier.
It used to accumulate binders from pi types indiscriminately, but this
led to unfriendly behavior. For example, `#check String.append` would
give
```
String.append (a✝ : String) (a✝¹ : String) : String
```
with inaccessible names. These appear because `String.append` is defined
using patterns, so it never names these parameters.
Now the delaborator stops accumulating binders once it reaches an
inaccessible name, and for example `#check String.append` now gives
```
String.append : String → String → String
```
We do not synthesize names for the sake of enabling binder syntax
because the binder names are part of the API of a function — one can use
`(arg := ...)` syntax to pass arguments by name. The delaborator also
now stops accumulating binders once it reaches a parameter with a name
already seen before — we then rely on the main delaborator to provide
that parameter with a fresh name when pretty printing the pi type.
As a special case, instance parameters with inaccessible names are
included as binders, pretty printing like `[LT α]`, rather than
relegating them (and all the remaining parameters) to after the colon.
It would be more accurate to pretty print this as `[inst✝ : LT α]`, but
we make the simplifying assumption that such instance parameters are
generally used via typeclass inference. Likely `inst✝` would not
directly appear in pretty printer output, and even if it appears in a
hover, users can likely figure out what is going on. (We may consider
making such `inst✝` variables pretty print as `‹LT α›` or
`infer_instance` in the future, to make this more consistent.)
Something we note here is that we do not do anything to make sure
parameters that can be used as named arguments actually appear named
after the colon (nor do we assure that the names are the correct names).
For example, one sees `foo : String → String → String` rather than `foo
: String → (baz : String) → String`. We can investigate this later if it
is wanted.
We also give `delabConstWithSignature` a `universes` flag to enable
turning off pretty printing universe levels parameters.
Closes#2846
the user can now write `termination_by?` to see the termination argument
inferred by GuessLex, and turn it into `termination_by …` using the “Try
this” widget or a code action.
To be done later, maybe: Avoid writing `sizeOf` if it's not necessary.
This PR addresses several performance issues in the auto-completion
implementation. It also fixes a number of smaller bugs related to
auto-completion.
In a file with `import Mathlib`, the performance of various kinds of
completions has improved as follows:
- Completing `C`: 49000ms -> 1400ms
- Completing `Cat`: 14300ms -> 1000ms
- Completing `x.` for `x : Nat`: 3700ms -> 220ms
- Completing `.` for an expected type of `Nat`: 11000ms -> 180ms
The following bugs have been fixed as well:
- VS Code never used our custom completion order. Now, the server fuzzy
completion score decides the order that completions appear in.
- Dot auto-completion for private types did not work at all. It does
now.
- Completing `.<identifier>` (where the expected type is used to infer
the namespace) did not filter by the expected type and instead displayed
all matching constants in the respective namespace. Now, it uses the
expected type for filtering. Note that this is not perfect because
sub-namespaces are technically correct completions as well (e.g.
`.Foo.foobar`). Implementing this is future work.
- Completing `.` was often not possible at all. Now, as long as the `.`
is not used in a bracket (where it may be used for the anonymous lambda
feature, e.g. `(. + 1)`), it triggers the correct completion.
- Fixes#3228.
- The auto-completion in `#check` commands would always try to complete
identifiers using the full declaration name (including namespaces) if it
could be resolved. Now it simply uses the identifier itself in case
users want to complete this identifier to another identifier.
## Details
Regarding completion performance, I have more ideas on how to improve it
further in the future.
Other changes:
- The feature that completions with a matching expected type are sorted
to the top of the server-side ordering was removed. This was never
enabled in VS Code because it would use its own completion item order
and when testing it I found it to be more confusing than useful.
- In the server-side ordering, we would always display keywords at the
top of the list. They are now displayed according to their fuzzy match
score as well.
The following approaches have been used to improve performance:
- Pretty-printing the type for every single completion made up a
significant amount of the time needed to compute the completions. We now
do not pretty-print the type for every single completion that is offered
to the user anymore. Instead, the language server now supports
`completionItem/resolve` requests to compute the type lazily when the
user selects a completion item.
- Note that we need to keep the amount of properties that we compute in
a resolve request to a minimum. When the server receives the resolve
request, the document state may have changed from the state it was in
when the initial auto-completion request was received. LSP doesn't tell
us when it will stop sending resolve requests, so we cannot keep this
state around, as we would have to keep it around forever.
LSP's solution for this dilemma is to have servers send all the state
they need to compute a response to a resolve request to the client as
part of the initial auto completion response (which then sends it back
as part of the resolve request), but this is clearly infeasible for all
real language servers where the amount of state needed to resolve a
request is massive.
This means that the only practical solution is to use the current state
to compute a response to the resolve request, which may yield an
incorrect result. This scenario can especially occur when using
LiveShare where the document is edited by another person while cycling
through available completions.
- Request handlers can now specify a "header caching handler" that is
called after elaborating the header of a file. Request handlers can use
this caching handler to compute caches for information stored in the
header. The auto-completion uses this to pre-compute non-blacklisted
imported declarations, which in turn allow us to iterate only over
non-blacklisted imported declarations where we would before iterate over
all declarations in the environment. This is significant because
blacklisted declarations make up about 4/5 of all declarations.
- Dot completion now looks up names modulo private prefixes to figure
out whether a declaration is in the namespace of the type to the left of
the dot instead of first stripping the private prefix from the name and
then comparing it. This has the benefit that we do not need to scan the
full name in most cases.
This PR also adds a couple of regression tests for fixed bugs, but *no
benchmarks*. We will add these in the future when we add proper support
for benchmarking server interaction sessions to our benchmarking
architecture.
All tests that were broken by producing different completion output
(empty `detail` field, added `sortText?` and `data?` fields) have been
manually checked by me to be still correct before replacing their
expected output.
This PR is an effort to improve reasoning at the Nat level about
bitvectors and reduce of Fin and Nat.
It slightly tightens some proofs, but is generally aimed at reducing
inconsistencies between definitions at the Nat and Fin types in favor of
more consistently using Nat operations.
This ports leanprover/std4#664 to Lean core.
Here was the rational I provided in the discussion for
leanprover/std4#664:
It's mostly about consistency. If we use the same types and style in
definitions and proofs, there is less surprise when unfolding or
otherwise using definitions. We use some Nat based operations that
haven't been extended to Fin such as the bitwise operations, and I don't
want to pay the overhead of introducing a Fin version of every Bitvector
operation.
So this basically means Nat is preferred.
One argument potentially in favor of Fin is that we could reuse results
proven there, but that doesn't really seem to be the case so far.
A second argument is that we want to simplify expression to use more
canonical forms and we currently can pretty-print those operations
better using ofNat than ofFin. We could define the notations using ofFin
of course though, but that's additional operators that will show up in
expressions.
This is not a complete upstreaming of that file (it also supports `∀ᵉ (x
< 2) (y < 3), p x y` as shorthand for `∀ x < 2, ∀ y < 3, p x y`, but I
don't think we need this; it is used in Mathlib).
Syntaxes still need to be made built-in.
---------
Co-authored-by: Leonardo de Moura <leomoura@amazon.com>
This change
* moves `termination_by` and `decreasing_by` next to the function they
apply to
* simplify the syntax of `termination_by`
* apply the `decreasing_by` goal to all goals at once, for better
interactive use.
See the section in `RELEASES.md` for more details and migration advise.
This is a hard breaking change, requiring developers to touch every
`termination_by` in their code base. We decided to still do it as a
hard-breaking change, because supporting both old and new syntax at the
same time would be non-trivial, and not save that much. Moreover, this
requires changes to some metaprograms that developers might have
written, and supporting both syntaxes at the same time would make
_their_ migration harder.
This makes hover info, go to definition, etc work for the `h` in `cases
h : e`. The implementation is similar to that used for the `generalize h
: e = x` tactic.
This PR adds per-package server options to resolve#2455. It is based on
the previous work in #2456, but takes a different approach: options are
loaded for the specific file in the file worker when `print-paths` is
called, instead of loading them in the watchdog with a separate Lake
command. This change addresses review comments made in #2456.
In doing so, it introduces two new Lake config fields:
- `leanOptions`: `-D` flag options that are passed to both the language
server and `lean` when building.
- `moreServerOptions`: `-D` flag options that are passed to the language
server.
Since `print-paths` must also accept a file path to compute the options
for that file, this PR is changing the API for `print-paths`. As there
have been numerous complaints about the name `print-paths`, I also
decided to change it to `setup-file` in this PR, since it would break
compatibility with the old Lake API anyways.
This PR deprecates the Lakefile field `moreServerArgs` in favor of
`moreGlobalServerArgs`, as suggested in the review for #2456.
Fixes#2455
---------
Co-authored-by: digama0 <mcarneir@andrew.cmu.edu>
This implements a request handler for the `textDocument/rename` LSP
request, enabling renames via F2. It handles both local renames (e.g.
`let x := 1; x` to `let y := 1; y`) as well as global renames
(definitions).
Unfortunately it does not work for "orphan" files outside a project, as
it uses ilean data for the current file and this does not seem to be
saved for orphan files. As a result, the test file does not work,
although one can manually test the implementation against a project such
as mathlib. (This issue already exists for the "references" request,
e.g. ctrl click on the first `x` in `let x := 1; x` takes you to the
second one only if you are not in an orphan file.)
* Fixesleanprover-community/mathlib4#7124
