| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Kernel should be mostly correct, higher levels do random stuff at
times.
|
|
This work makes it possible to take advantage of a compact
representation for integers in the entire system, as opposed to only
in some reduction machines. It is useful for heavily computational
applications, where even constructing terms is not possible without such
a representation.
Concretely, it replaces part of the retroknowledge machinery with
a primitive construction for integers in terms, and introduces a kind of
FFI which maps constants to operators (on integers). Properties of these
operators are expressed as explicit axioms, whereas they were hidden in
the retroknowledge-based approach.
This has been presented at the Coq workshop and some Coq Working Groups,
and has been used by various groups for STM trace checking,
computational analysis, etc.
Contributions by Guillaume Bertholon and Pierre Roux <Pierre.Roux@onera.fr>
Co-authored-by: Benjamin Grégoire <Benjamin.Gregoire@inria.fr>
Co-authored-by: Vincent Laporte <Vincent.Laporte@fondation-inria.fr>
|
|
|
|
|
|
|
|
|
|
|
|
This is to move a standard combinator to the place it belongs to. An
alternative could have been to put it in termops.ml, but termops.ml is
now about econstr, so, even if it makes the kernel "bigger", constr.ml
seems to be the best place for this combinator. After all, this
combinator is canonical.
|
|
A few of them will be of help for future cleanups. We have spared the
stuff in `Names` due to bad organization of this module following the
split from `Term`, which really difficult things removing the
constructors.
|
|
|
|
|
|
|
|
We build the `@doc` target in the `dune` job:
- The documentation can be found in `_build/default/_doc/`
- We had to fix a couple of quoting problems.
|
|
More precisely: the lambda-let-expanded canonical form of branches and
return predicate is considered as part of the structure of a "match"
and is preserved.
|
|
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).
|
|
And similarly for fixpoint and cofixpoint.
|
|
Dependency analysis for separate compilation was not iterated properly
on rel_context and named_context.
|
|
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.
|
|
|
|
Previously [fun x : Ind@{i} => x : Ind@{j}] with Ind some cumulative
inductive would try to generate a constraint [i = j] and use
cumulativity only if this resulted in an inconsistency. This is
confusingly different from the behaviour with [Type] and means
cumulativity can only be used to lift between universes related by
strict inequalities. (This isn't a kernel restriction so there might
be some workaround to send the kernel the right constraints, but
not in a nice way.)
See modified test for more details of what is now possible.
Technical notes:
When universe constraints were inferred by comparing the shape of
terms without reduction, cumulativity was not used and so too-strict
equality constraints were generated. Then in order to use cumulativity
we had to make this comparison fail to fall back to full conversion.
When unifiying 2 instances of a cumulative inductive type, if there
are any Irrelevant universes we try to unify them if they are
flexible.
|
|
|
|
We separate functions dealing with VM values (vmvalues.ml) and
interfaces of the bytecode interpreter (vm.ml). Only the former relies
on untyped constructions.
This also makes the VM architecture closer to the one of native_compute,
another patch could probably try to share more code between the two for
conversion and reification (not trivial, though).
This is also preliminary work for integers and arrays.
|
|
The comment had the type and value of the let-in swapped, which contradicted the listed types.
|
|
There don't really bring anything, we also correct some minor nits
with the printing function.
|
|
We mirror the structure of EConstr and move the destructors from `Term`
to `Constr`.
This is a step towards having a single module for `Constr`.
|
|
We do up to `Term` which is the main bulk of the changes.
|
|
This will allow to merge back `Names` with `API.Names`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1. The Univ module now only cares about definitions about universes.
2. The UGraph module contains the algorithm responsible for aciclicity.
|
|
Followup of: f7b29094fe7cc13ea475447bd30d9a8b942f0fef . In particular, re-closes #3593.
As a side effect, fixes an undiscovered bug of the `eq_constr` tactic which didn't consider terms up to evar instantiation.
|
|
[kind_of_term].
To be able to write equality up to evar instantiation instantiation.
Generalises the main function of [eq] constr over the variant of [kind_of_term] it uses. It prevents some optimisation of [Array.equal] where two physically equal arrays are considered (less or) equal. But it does not seem to have appreciable effects on efficiency.
|
|
[compare_head_gen] defined in terms of [compare_head_gen_leq]. Remove an unused argument from [compare_head_gen_leq].
|
|
|
|
but the internal representation dropped let-in.
Ideally, the internal representation of the "match" should use
contexts for the predicate and the branches. This would however be a
rather significant change. In the meantime, just a hack.
To do, there is still an extra @ in the constructor name that does not
need to be there.
|
|
so as to reproduce correctly the reduction behavior of existing
projections, i.e. delta + iota. Make [projection] an abstract datatype
in Names.ml, most of the patch is about using that abstraction.
Fix unification.ml which tried canonical projections too early in
presence of primitive projections.
|
|
Universes.
Needed to exponse compare_head_gen(_leq) so that it could be reused in Universes.
Remove unused functions from univ as well and refactor a little bit.
Changed the syntax to Type@{} for explicit universe level specs, following the WG decision.
|
|
allowing fast conversion to be used during unification while respecting the
semantics of unification w.r.t universes.
- Inside kernel, checked_conv is used mainly, it just does checking, while infer_conv
is used for module subtyping.
- Outside, infer_conv is wrapped in Reductionops to register the right constraints
in an evarmap.
- In univ, add a flag to universes to cache the fact that they are >= Set, the
most common constraints, resulting in an 4x speedup in some cases (e.g. HigmanS).
|
|
needed during
unification.
|
