| Age | Commit message (Collapse) | Author |
|---|
|
|
|
pattern-matching predicate.
|
|
|
|
|
|
types: we downcast the evar in the higher type to the lower type.
Then, we have the freedom to choose the order of instantiation
according to the instances of the evars (e.g. choosing the
instantiation for which pattern-unification is possible, if ever it is
possible in only one way - as shown by an example in contribution
Paco).
This still does not preserve compatibility because it happens that
type classes resolution is crucially dependent on the order of
presentation of the ?n=?p problems. Consider e.g. an example taken
from Containers. Both now and before e2fa65fccb9, one has this asymmetry:
Context `{Helt : OrderedType elt}.
Check forall x y r l h, @Equivalence.equiv _ _ _ x y -> In x (Node l x r h).
--> @Equivalence.equiv QArith_base.Q QArith_base.Qeq QArith_base.Q_Setoid x y
Context `{Helt : OrderedType elt}.
Check forall x y r l h, In x (Node l x r h) -> @Equivalence.equiv _ _ _ x y.
--> @Equivalence.equiv elt (@_eq elt Helt) (@OT_Equivalence elt Helt)
Then, embedding this example within a bigger one which relies on the
?n=?n' resolution order, one can get two incompatible resolution of
the same problem.
To be continued...
|
|
inductive types + deactivate test for equality of sort + deactivate
the check that the constraints Prop/Set <= Type are declared).
|
|
- drops all Defined entries from the evar map (applying the subst to the
initial evar and the undefined evars types).
- call Gc.compact
Now the question is: where should these two commands be documented?
|
|
One missing evar was making the whole substitution fail.
The bug actually existed a priori also in the case where only metas
are substituted (i.e. before bcba6d1bc9f769da), leading to limit the
number of situations where it could be effectful.
This fixes current failures of RelationAlgebra and CFGV.
|
|
eager meta substition in w_unify, so as to preserve compatibility
after PMP's move of (setoid) rewrite clauses from metas to
evars (fbbe491cfa).
Hoping it is compatible for non-rewrite uses of the eager meta flag,
and that it is not too costly.
|
|
|
|
|
|
Its semantics was dubious, and it was not used anymore anyway.
|
|
The main change is that selection of subterm is made similar whether
the given term is fully applied or not.
- The selection of subterm now works as follows depending on whether
the "at" is given, of whether the subterm is fully applied or not,
and whether there are incompatible subterms matching the pattern. In
particular, we have:
"at" given
| subterm fully applied
| | incompatible subterms
| | |
Y Y - it works like in 8.4
Y N - this was broken in 8.4 ("at" was ineffective and it was finding
all subterms syntactically equal to the first one which matches)
N Y Y it now finds all subterms like the first one which matches
while in 8.4 it used to fail (I hope it is not a too risky in-draft
for a semantics we would regret...) (e.g. "destruct (S _)" on
goal "S x = S y + S x" now selects the two occurrences of "S x"
while it was failing before)
N Y N it works like in 8.4
N N - it works like in 8.4, selecting all subterms like the
first one which matches
- Note that the "historical" semantics, when looking for a subterm, to
select all subterms that syntactically match the first subterm to
match the pattern (looking from left to right) is now internally called
"like first".
- Selection of subterms can now find the type by pattern-matching (useful e.g.
for "induction (nat_rect _ _ _ _)")
- A version of Unification.w_unify w/o any conversion is used for
finding the subterm: it could be easily replaced by an other
matching algorithm.
In particular, "destruct H" now works on a goal such as "H:True -> x<=y |- P y".
Secondary change is in the interpretation of terms with existential
variables:
- When several arguments are given, interpretation is delayed at the
time of execution
- Because we aim at eventually accepting "edestruct c" with unresolved
holes in c, we need the sigma obtained from c to be an extension of
the sigma of the tactics, while before, we just type-checked c
independently of the sigma of the tactic
- Finishing the resolution of evars (using type classes, candidates,
pending conversion problems) is made slightly cleaner: it now takes
three states: a term is evaluated in state sigma, leading to state
sigma' >= sigma, with evars finally solved in state sigma'' >=
sigma'; we solve evars in the diff of sigma' and sigma and report
the solution in sigma''
- We however renounce to give now a success semantics to "edestruct c"
when "c" has unresolved holes, waiting instead for a decision on
what to do in the case of a similar eapply (see mail to coqdev).
An auxiliary change is that an "in" clause can be attached to each component
of a "destruct t, u, v", etc.
Incidentally, make_abstraction does not do evar resolution itself any longer.
|
|
For optimisation purposes.
|
|
|
|
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.
|
|
|
|
|
|
another one.
|
|
and unsatisfiable constraints which were not done in the right environment.
|
|
|
|
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.
|
|
(the action is "clear").
Added subst_intropattern which was missing since the introduction of
ApplyOn intro patterns.
Still to do: make "intros _ ?id" working without interferences when
"id" is precisely the internal name used for hypotheses to discard.
|
|
will name the goal id; writing ?[?id] will use the first
fresh name available based with prefix id.
Tactics intro, rename, change, ... from logic.ml now preserve goal
name; cut preserves goal name on its main premise.
|
|
- Removed collect_evars which does not consider instance
(use evars_of_term instead).
- Also removed evars_of_evar_info which did not filter context (use
evars_of_filterered_evar_info instead). This is consistent with
printing goal contexts in the filtered way.
Anyway, as of today, afaics goals filters are trivial
because (if I interpret evarutil.ml correctly), evars with
non-trivial filter necessarily occur in a conv pb. Conversely,
conv pbs being solved when tactics are called, there should not be
an evar used as a goal with a non-trivial filter.
|
|
with existing ML code.
|
|
|
|
equality of universes, along with a few other functions in evd.
|
|
|
|
|
|
|
|
in instances.
|
|
|
|
instances still to do). Using heuristics to name after the quantifier
name it comes. Also added a "sigma" to almost all printing functions.
|
|
for typeclass errors.
|
|
|
|
Names and arguments were uniformized, and some functions were redesigned to
take their typical use-case into account.
|
|
The levels added by the context are in general much fewer than the size of the
evarmap, so it is better to add them to the latter rather than doing it the
other way around.
|
|
|
|
unification, speeding it up considerably
Revert backwards-incompatible commit 77df7b1283940d979d3e14893d151bc544f41410
|
|
of the universe context in the obligations, it gets gradually fixed
globally by each one of them.
Fixes bug found in Misc/Overloading.
|
|
|
|
variable
as algebraic so it can disappear from the proof (it always gets substituted away from
the term). This means less spurious universes remaining in proof terms.
|
|
|
|
for helping fixing this).
Now the issue is handled solely through refreshing of the terms assigned
to evars during unification.
If ?X = list ?Y, then Y's type is refreshed so that it doesn't mention
a template universe and in turn, ?X won't. Same goes when typechecking
(nil ?X, nil ?Y), the pair constructor levels will be set higher than fresh
universes for the lists carriers. This also handles user-defined functions
on template polymorphic inductives, which was fragile before.
Pretyping and Evd are now uncluttered from template-specific code.
|
|
and avoiding explicit substitutions and merging of contexts, e.g. in obligations.ml.
The context produced by typechecking a statement is passed in the proof, allowing the
universe name context to be correctly folded as well. Mainly an API cleanup.
|
|
|
|
|
