| Age | Commit message (Collapse) | Author |
|---|
|
It is the only place where it starts making sense in the whole codebase. It also
fits nicely there since there are other functions manipulating this type in that
module.
In any case this type does not belong to the kernel.
|
|
In our quest to unify all the declaration paths, an important step
is to account for the state pertaining to `Program` declarations.
Whereas regular proofs keep are kept in a stack-like structure;
obligations for constants defined by `Program` are stored in a global
map which is manipulated by almost regular open/close proof primitives.
We make this manipulation explicit by handling the program state
functionally, in a similar way than we already do for lemmas.
This requires to extend the proof DSL a bit; but IMO changes are
acceptable given the gain.
Most of the PR is routine; only remarkable change is that the hook is
called explicitly in `finish_admitted` as it had to learn about the
different types of proof_endings.
Note that we could have gone deeper and use the type system to refine
the core proof type; IMO it is still too preliminary so it is better
to do this step as an intermediate one towards a deeper unification.
|
|
At this point the record in lemmas was just a stub; next commit will
stop exposing the internals of mutual information, and pave the way
for the refactoring of `Info.t` handling in the Declare interface.
|
|
The API in `DeclareDef` should become the recommended API in `Declare`.
This greatly reduces the exposure of internals; we still have a large
offender in `Lemmas` but that will be taken care of in the next
commit; effectively removing quite some chunks from `declare.mli`.
This PR originally introduced a dependency cycle due to:
- `Declare`: uses `Vernacexpr.decl_notation list`
- `Vernacexpr`: uses `ComHint.hint_expr`
- `ComHint`: uses `Declare.declare_constant`
This is a real cycle in the sense that `ComHint` would have also move
to `DeclareDef` in the medium term.
There were quite a few ways to solve it, we have chosen to
move the hints ast to `Vernacexpr` as it is not very invasive
and seems consistent with the current style.
Alternatives, which could be considered at a later stage are for
example moving the notations AST to `Metasyntax`, having `Declare` not
to depend on `Vernacexpr` [which seems actually a good thing to do in
the medium term], reworking notation support more deeply...
|
|
This moves the vernacular part of hints to `vernac`; in particular, it
helps removing the declaration of constants as parts of the `tactic`
folder.
|
|
Add headers to a few files which were missing them.
|
|
We can now do `#[refine] Instance : Bla := bli.` to enter proof mode
with `bli` as a starting refinement.
If `bli` is enough to define the instance we still enter proof mode,
keeping things nicely predictable for the stm.
|
|
This is more in-line with attributes and the rest of the API, and
makes some code significantly clearer (as in `foo true false false`,
etc...)
|
|
|
|
This replaces the mismatched context error, which occurred when
Instance := {} was used with strictly more fields than declared.
Since we later check that field names match those declared for the
instance, now that we reject duplicates we know that there are no
extra fields.
|
|
The main idea of this PR is to distinguish the types of "proof object"
`Proof_global.t` and the type of "proof object associated to a
constant, the new `Lemmas.t`.
This way, we can move the terminator setup to the higher layer in
`vernac`, which is the one that really knows about constants, paving
the way for further simplification and in particular for a unified
handling of constant saving by removal of the control inversion here.
Terminators are now internal to `Lemmas`, as it is the only part of
the code applying them.
As a consequence, proof nesting is now handled by `Lemmas`, and
`Proof_global.t` is just a single `Proof.t` plus some environmental
meta-data.
We are also enable considerable simplification in a future PR, as this
patch makes `Proof.t` and `Proof_global.t` essentially the same, so we
should expect to handle them under a unified interface.
|
|
deprecated attribute
This code was significantly more complex than necessary.
|
|
|
|
|
|
|
|
![proof_stack] is equivalent to the old meaning of ![proof]: the body
has type `pstate:Proof_global.t option -> Proof_global.t option`
The other specifiers are for the following body types:
~~~
![open_proof] `is_ontop:bool -> pstate`
![maybe_open_proof] `is_ontop:bool -> pstate option`
![proof] `pstate:pstate -> pstate`
![proof_opt_query] `pstate:pstate option -> unit`
![proof_query] `pstate:pstate -> unit`
~~~
The `is_ontop` is only used for the warning message when declaring a
section variable inside a proof, we could also just stop warning.
The specifiers look closely related to stm classifiers, but currently
they're unconnected. Notably this means that a ![proof_query] doesn't
have to be classified QUERY.
![proof_stack] is only used by g_rewrite/rewrite whose behaviour I
don't fully understand, maybe we can drop it in the future.
For compat we may want to consider keeping ![proof] with its old
meaning and using some new name for the new meaning. OTOH fixing
plugins to be stricter is easier if we change it as the errors tell us
where it's used.
|
|
Typically instead of [start_proof : ontop:Proof_global.t option -> bla ->
Proof_global.t] we have [start_proof : bla -> Proof_global.pstate] and
the pstate is pushed on the stack by a caller around the
vernacentries/mlg level.
Naming can be a bit awkward, hopefully it can be improved (maybe in a
followup PR).
We can see some patterns appear waiting for nicer combinators, eg in
mlg we often only want to work with the current proof, not the stack.
Behaviour should be similar modulo bugs, let's see what CI says.
|
|
It's used a few times in the stdlib (a couple of which need no other
change when removing the !) and not at all throughout our CI.
Considering that I think it's fair enough to remove it.
|
|
|
|
I had to reorganize the code a bit. The Context command moved to
comAssumption, as it is not so related to type classes. We were able to
remove a few hooks on the way.
|
|
Not sure the warning is worth the extra parameter.
|
|
|
|
We remove all calls to `Flags.is_program_mode` except one (to compute
the default value of the attribute). Everything else is passed
explicitely, and we remove the special logic in the interpretation loop
to set/unset the flag.
This is especially important since the value of the flag has an impact on
proof modes, so on the separation of parsing and execution phases.
|
|
This makes code paths clearer (we still factorize a lot of the
treatment), and we seize the opportunity to forbid anonymous
`Declare Instance` which is not a documented construction, and seems to
make little sense in practice.
|
|
This is a pre-requisite to use automated formatting tools such as
`ocamlformat`, also, there were quite a few places where the comments
had basically no effect, thus it was confusing for the developer.
p.s: Reading some comments was a lot of fun :)
|
|
`Classes.new_instance` is one of the largest functions of the
codebase; we split it up and reduce indentation.
This will help further cleanups. This PR should introduce no code
changes other than splitting the function up.
|
|
This shall eventually allow to use contexts of declarations in the
definition of the "Case" constructor.
Basically, this means that Constr now includes Context and that the
"t" types of Context which were specialized on constr are not defined
in Constr (unfortunately using a heavy boilerplate).
|
|
reference was defined as Ident or Qualid, but the qualid type already
permits empty paths. So we had effectively two representations for
unqualified names, that were not seen as equal by eq_reference.
We remove the reference type and replace its uses by qualid.
|
|
Continuing the interface cleanup we place `Constrexpr` in the
internalization module, which is the one that eliminates it.
This slims down `pretyping` considerably, including removing the
`Univdecls` module which existed only due to bad dependency ordering
in the first place. Thanks to @ Skyskimmer we also remove a duplicate
`univ_decl` definition among `Misctypes` and `UState`.
This is mostly a proof of concept yet as it depends on quite a few
patches of the tree. For sure some tweaks will be necessary, but it
should be good for review now.
IMO the tree is now in a state where we can could easy eliminate more
than 10 modules without any impact, IMHO this is a net saving API-wise
and would help people to understand the structure of the code better.
|
|
Close #7562.
[api] move hint_info ast to tactics.
|
|
In #6092, `global_reference` was moved to `kernel`. It makes sense to
go further and use the current kernel style for names.
This has a good effect on the dependency graph, as some core modules
don't depend on library anymore.
A question about providing equality for the GloRef module remains, as
there are two different notions of equality for constants. In that
sense, `KerPair` seems suspicious and at some point it should be
looked at.
|
|
`hint_info_expr`, `hint_info_gen` do conceptually belong to the
typeclasses modules and should be able to be used without a dependency
on the concrete vernacular syntax.
|
|
|
|
This is useful as it allows to reflect program_mode behavior as an attribute.
|
|
We do up to `Term` which is the main bulk of the changes.
|
|
Introduce a "+" modifier for universe and constraint declarations to
indicate that these can be extended in the final definition/proof. By
default [Definition f] is equivalent to [Definition f@{+|+}], i.e
universes can be introduced and constraints as well. For [f@{}] or
[f@{i j}], the constraints can be extended, no universe introduced, to
maintain compatibility with existing developments. Use [f@{i j | }] to
indicate that no constraint (nor universe) can be introduced. These
kind of definitions could benefit from asynchronous processing.
Declarations of universe binders and constraints also works for
monomorphic definitions.
|
|
|
|
This is a bit long, but it is to keep a symmetry with constr_expr.
|
|
Currently, the STM, vernac interpretation, and the toplevel are
intertwined in a mutual dependency that needs to be resolved using
imperative callbacks.
This is problematic for a few reasons, in particular it makes the
interpretation of commands that affect the document quite intricate.
As a first step, we split the `toplevel/` directory into two: "pure"
vernac interpretation is moved to the `vernac/` directory, on which
the STM relies.
Test suite passes, and only one command seems to be disabled with this
approach, "Show Script" which is to my understanding
obsolete. Subsequent commits will fix this and refine some of the
invariants that are not needed anymore.
|
