| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
By moving convert_concl to new proof engine, re-typecheckeing was
incidentally introduced. Re-typechecking revealed that vm bug #2729
was still open. Indeed, the vm was still producing an ill-typed term.
This commit fixes ill-typedness of the vm in #2729 when reconstructing
a "match" in an inductive type whose constructors have let-ins.
Another part is still open (see test-suite).
|
|
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.
|
|
|
|
early call the standard resolution function which e.g. does
restriction and maybe solve the problem if pattern-like, instead of
postponing the problem.
|
|
don't print bindings of the form "x:=x" unless there is also a binding
"x':=x". Otherwise said, if a variable occurs several time, the
binding where it occurs under the form "x:=x" is printed anyway. This
should help to understand where the instance is non trivial while
still not obfuscating display in goals with very long list of
uninteresting trivial instances.
|
|
|
|
In my first attempt I just dropped all future goals before starting a refinement. This was done for simplicity but is incorrect in general. In this version the future goals which are not introduced by the particular instance of refine are kept for future use.
|
|
See previous commit for more discussion.
Changed the name from "main" to "principal" because I find "main" overused, and because the name was only introduced yesterday anyway.
|
|
That way, everything in the code of pretying is made "refine"-aware. Making the abstraction stonger and integration of pretyping with interactive proof more direct.
It might create goals in a slightly different goal order in the (user level) refine tactic. Because now, the [update] primitive which used to infer an order from an [evar_map] now has the order fixed by the successive declaration with [Evarutil.new_evar] (and similar). It probably coincides, though.
Following a suggestion by Hugo.
|
|
|
|
|
|
projections in cbv when delta _and_ beta flags are set. Add test-suite
file for bug 3700 too.
|
|
primitive record.
|
|
|
|
forms in evarconv and unification, as well as fallback to first-order
unification when eta for constructors fail. Update test-suite file
3484 to test for the FO case in evarconv as well.
|
|
folded primitive projections in applicative stacks in rhs as named, hence
prefering to unfold the lhs in these cases.
|
|
required, i.e. in first-order unification cases where the head of the
other side is a hole or the eta-expanded constant.
|
|
make printing exponentially slower. We would have to expand all projections
at once before detyping to make this linear.
|
|
|
|
unification.
|
|
another one.
|
|
|
|
|
|
|
|
|
|
for the record binder of classes. This name is no longer generated
in the kernel but part of the declaration. Also cleanup the interface
to recognize primitive records based on an option type instead of a
dynamic check of the length of an array.
|
|
in cases of unification with existentials requiring it.
|
|
and unsatisfiable constraints which were not done in the right environment.
|
|
application case to expand primitive projections at the head of both
applications.
|
|
expand_projection failing if an innapropriate sigma is given.
|
|
|
|
|
|
2nd-order matching).
We made the filter type-safe (i.e. to ensure that Gamma |- ?n:T is
well-typed when taking the filtered context Gamma) in 2nd order
matching. Maybe this weakens the power of the 2nd order matching
algorithm, so it is not clear that it is the good fix.
Another fix could have been to ensure that taking the closure of
filter does not extend it beyond the original filter (hence keeping
the filter untyped if it was already untyped).
As for the bug #3045, it also revealed that some "destruct c as [[]]"
could be made stronger as decomposing the destruct in two parts
"destruct c as [x]; destruct x" workis while it currently fails if in
one part.
|
|
to maintain compatibility (HoTT bug #557).
|
|
|
|
- Optimize the removal of generalization when there is no dependency in
the generalized variable (see postprocess_dependencies, and the removal
of dependencies in the default type of impossible cases).
- Compute the onlydflt flag correctly (what allows automatic treatment
of impossible cases even when there is no clause at all).
|
|
|
|
|
|
|
|
scope" error message).
|
|
it became possible to have binding names themselves bound to ltac
variables (2fcc458af16b).
|
|
glob_constr to constr_pattern. Was partially fixed to solve #3088
(8e88b7adab) in but the order of lambdas was still incorrect as the
fix of the order of lambdas in second-order pattern-matching for #3136
showed (83159124ce22).
|
|
potentially different types, resulting in ill-typed terms due to eta.
Projection expansion now fails gracefully on retyping errors.
The proper fix to unification, checking that the heads for FO
have unifiable types, is currently too strong, adding unnecessary universe
constraints, so it is disabled for now. It might be quite expensive
too also it's not noticeable on the stdlib.
|
|
|
|
|
|
|
|
by names): VarInstance behaves like GoalEvar for type class
resolution.
|
|
reorganization of apply in d5fece25d8964d5d9fcd55b66164286aeef5fb9f:
using renaming also in retyping.
|
