| Age | Commit message (Collapse) | Author |
|---|
|
non-dependent or propositional constraint has already been found
(same behavior as before previous patch).
|
|
Add a global option to check for multiple solutions and fail in that
case.
Add another flag to declare classes as having unique instances (not
checked but assumed correct, avoiding some backtracking).
|
|
|
|
for tclEVARS which might solve existing goals.
|
|
was working in 8.4.
|
|
|
|
generated equation names).
|
|
was failing in this case due to the wrong postponment of an unsolvable
?X = RigidContext[?X] problem.
|
|
|
|
When w_unifying primitive projection applications, force
the unification of types of the projected records to recover instances
for the parameters (evarconv does this automatically by unifying evar
instances with their expected type).
|
|
implicits do not allow to parse as an application and cleanup code.
|
|
in69665dd2480d364162933972de7ffa955eccab4d. There are still situations
when "as" is not given where equations coming from injection are not
yet removed, making invalid the computation of dependencies, what
prevents an hypothesis to be cleared and replaced.
|
|
|
|
Printing All cases (bug #3597).
- Fix Ltac matching with primitive projections (bug #3598).
- Spotted a problem with printing of constants with maximally implicit
arguments due to strange "compatibility" interpretation of Arguments [X]
as Arguments {X} but didn't fix it entirely yet (might cause incompatibilities).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
use of projections.
|
|
|
|
|
|
|
|
|
|
|
|
introducing constants (e.g. Top.1 is always before Top.2), compatible with the one
used before introduction of hMaps in LMap/LSet.
|
|
|
|
|
|
and their eta-expanded forms succeed, potentially filling parameter
metavariables from the type information of the projected object.
|
|
|
|
|
|
|
|
fresh names interferring with names later generated in intropatterns
(as introduced in 72498d6d68ac which passed "clenv_refine_in continued
by pattern introduction" to descend_in_conjunction while only a pure
clenv_refine was passed before). The 72498d6d68ac version was
generating fresh names in the wrong order (morally-private names for
descend_in_conjunction before user-level names in clenv_refine_in).
The 19a7a5789c fix was introducing expressions not handled by the
refine from logic.ml.
In particular, this fixes RelationAlgebra.
|
|
|
|
now done entirely using declare_mind, which declares the associated
constants for primitive records. This avoids a hack related to
elimination schemes and ensures that the forward references to constants
in the mutual inductive entry are properly declared just after the
inductive. This also clarifies (and simplifies) the code of term_typing
for constants which does not have to deal with building
or checking projections anymore.
Also fix printing of universes showing the de Bruijn encoding in a few places.
|
|
at least remove the successes obtained by trivial unification of a
meta with the goal, so as to avoid surprising results. We generalize
this to variables which will only eventually be replaced by metas.
|
|
eta-expanded version of a projection as before.
|
|
The unification oracle now prefers unfolding the eta-expanded form of a
projection over the primitive projection, and allows first-order
unification on applications of constructors of primitive records,
in case eta-conversion fails (disabled by previous patch on eta).
|
|
pruning),
hence do not entirely prevent solve_simple_eqn in case of apparent occurs-check but
backtrack to eqappr on OccurCheck failures (problem found in Ssreflect).
|
|
presence of let-ins.
|
|
|
|
|
|
to match on a primitive projection application c.(p) using "?f _", binding f
to (fun x => x.(p)) with the correct typing.
|
|
|
|
It's possible that I should have removed more "allows", as many
instances of "foo allows to bar" could have been replaced by "foo bars"
(e.g., "[Qed] allows to check and save a complete proof term" could be
"[Qed] checks and saves a complete proof term"), but not always (e.g.,
"the optional argument allows to ignore universe polymorphism" should
not be "the optional argument ignores universe polymorphism" but "the
optional argument allows the caller to instruct Coq to ignore universe
polymorphism" or something similar).
|
|
|
|
|
|
|
|
|
