| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
| |
resolution of generated evars, not doing any backtracking yet.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12175 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
as they are used.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12148 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
| |
- Set implicit args on for Context decls
- Move class_apply tactic to Init
- Normalize evars before raising an error.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12127 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Decorate proof search with depth/subgoal number information
- Add a tactic [autoapply foo using hints] to call the resolution tactic
with the [transparent_state] associated with a hint db, giving better
control over unfolding.
- Fix a bug in the Lambda case in the new rewrite
- More work on the [Proper] and [subrelation] hints to cut the search space
while retaining completeness.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12118 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
- [matches] is not parameterized by evars: normalize before calling
conclPattern.
- fix hints in Morphisms for subrelation and handling of signature
normalization.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12115 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
| |
and failure continuations, allowing to do safe cuts correctly.
- Fix bug #2097 by suppressing useless nf_evars calls.
- Improve the proof search strategy used by rewrite for subrelations and
fix some hints.
Up to 20% speed improvement in setoid-intensive files.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12110 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
| |
clause resulting in stray notations for e.g. variable named "le")
and 12083 (fixing bug in as clause of apply in) from trunk.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12103 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
with standard math nomenclature. Also clean up in rewrite.ml4.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12097 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
Setoid doesn't export Program.Basics anymore.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12095 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
| |
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12094 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
Setoid, the rest needs to be qualified.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12093 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
Program.Tactics anymore.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12089 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
strategies (à la ELAN).
Now setoid_rewrite is just one strategy and autorewrite is a
more elaborate one. Any kind of traversals can be defined, strategies
can be composed etc... in ML. An incomplete (no fix) language extension
for specifying them in Ltac is there too.
On a typical autorewrite-with-setoids usage, proof production time is
divided by 2 and proof size by 10.
Fix some admitted proofs and buggy patterns in Classes.Morphisms as
well.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12081 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
class_tactics.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12056 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- The measure can now refer to all the formal arguments
- The recursive calls can make all the arguments vary as well
- Generalized to any relation and measure (new syntax {measure m on R})
This relies on an automatic curryfication transformation, the real
fixpoint combinator is working on a sigma type of the arguments.
Reduces to the previous impl in case only one argument is involved.
The patch also introduces a new flag on implicit arguments that says if
the argument has to be infered (default) or can be turned into a
subgoal/obligation. Comes with a test-suite file.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12030 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
for [PartialOrder] typeclass.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11882 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
| |
- curly braces mandatory for record class instances
- no mention of the unique method for definitional class instances
Turning a record or definition into a class is mostly a
matter of changing the keywords to 'Class' and 'Instance' now.
Documentation reflects these changes, and was checked once more
along with setoid_rewrite's and Program's.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11797 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
doc is ok). Rework the .v files in Program accordingly, adding some
documentation and proper headers. Integrate the development of an
elimination principle for measured functions in Program/Wf by Eelis van
der Weegen.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11686 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
variant of the [unify] tactic that takes a hint db as argument and does
unification modulo its [transparent_state]. Add test-file for bug #1939
and another [AdvancedTypeClasses.v] that mimicks
[AdvancedCanonicalStructure.v].
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11685 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
guessing the binding name by default and making all generalized
variables implicit. At the same time, continue refactoring of
Record/Class/Inductive etc.., getting rid of [VernacRecord]
definitively. The AST is not completely satisfying, but leaning towards
Record/Class as restrictions of inductive (Arnaud, anyone ?).
Now, [Class] declaration bodies are either of the form [meth : type] or
[{ meth : type ; ... }], distinguishing singleton "definitional" classes
and inductive classes based on records. The constructor syntax is
accepted ([meth1 : type1 | meth1 : type2]) but raises an error
immediately, as support for defining a class by a general inductive type
is not there yet (this is a bugfix!).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11679 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
procedure. Brainstormed with Nicolas Tabareau.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11660 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
for the report).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11656 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
| |
declaration code to toplevel/record, including support for singleton
classes as definitions. Parsing code also factorized. Arnaud: one more
thing to think about when refactoring the definitions in vernacentries.
Add support for specifying what to do with anonymous variables in
contexts during internalisation (fixes bug #1982), current choice is to
generate a name for typeclass bindings.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11563 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
| |
- Rework definition of the type of respecful functions in Morphisms.v
- Unfold [flip] in "Add Morphism" tactic (suggested by N. Tabareau)
- Add a "coqvariableref" command in coqdoc.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11545 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
instead of a completely resolved [constr] as input. The propagation of the
evars associated to the lemma is only allowed when the evar flag is on
(i.e. for [eapply]), otherwise they should be resolved by the end of the
application. Should be completely backwards compatible...
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11497 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- New constr_expr construct [CGeneralization of loc * binding_kind *
abstraction_kind option * constr_expr] to generalize the free vars of
the [constr_expr], binding these using [binding_kind] and making
a lambda or a pi (or deciding from the scope) using [abstraction_kind
option] (abstraction_kind = AbsLambda | AbsPi)
- Concrete syntax "`( a = 0 )" for explicit binding of [a] and "`{
... }" for implicit bindings (both "..(" and "_(" seem much more
difficult to implement). Subject to discussion! A few examples added
in a test-suite file.
- Also add missing syntax for implicit/explicit combinations for
_binders_: "{( )}" means implicit for the generalized (outer) vars,
explicit for the (inner) variable itself. Subject to discussion as well :)
- Factor much typeclass instance declaration code. We now just have to
force generalization of the term after the : in instance declarations.
One more step to using Instance for records.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11495 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
| |
[discriminated]] with a switch for using the more experimantal dnet impl
for every hint. Also add [Hint Transparent/Opaque] which parameterize
the dnet and the unification flags used by auto/eauto with a particular
database. Document all this. Remove [Typeclasses unfold] directives that
are no longer needed (everything is unfoldable by default) and move to
[Typeclasses Transparent/Opaque] syntax.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11409 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
Now it's [A] and [R] for carriers and relations and [eqA] when the
relation is supposed to be an equivalence. The types are always implicit
except for [pointwise_relation] which now takes the domain type as an
explicit argument.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11408 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
| |
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11401 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
| |
recursive definitions and references to previous fields in record and
classes definitions. Fixes the corresponding typesetting issue in coqdoc
output.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11397 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This permits to create a database [relations] in [RelationClasses] with
a single extern tactic in it that tries to apply [reflexivity] or
[symmetry]. This is then automatically used in [auto with *] and repair
backward compatibility. The previous commit broke some scripts which were
using [intuition] to do (setoid) [reflexivity] or [symmetry]: this
worked only by accident, because the hint database of typeclasses was
used. Overrall, this also allows to put a bunch of
always-applicable, related tactics in some database or to use
[Hint Extern] but match only on hypotheses.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11384 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
only relevant hypotheses in it, slightly cutting the search space.
Makes setoid_rewrite less sensitive to the number of hypothesis (~ 5%
time improvment on the stdlib). This currently prevents resolution with
constants which reduce to classes, hence the modifications in Setoids.Setoid.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11360 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
| |
the same typeclass method application tactic that's available to users.
Modify a bit the _red tactics to accomodate the new setup and comment
some dead code in setoid_replace. Next step is removing it completely.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11355 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Avoid using projections for singleton classes. Divides the size
of proofs by 2 and simplifies the typechecking as well.
- Switch to the new discrimination net implementation for classes,
with the current convention that all constants are unfoldable.
Users can add "Typeclasses Opaque" declarations make the dnet
discriminate more, otherwise it should be entirely
backward-compatible.
- Fix bug introduced in r11333 in "transitivity" tactic in presence
of coercions.
Up to 15% gains on setoid-intensive files like Ring_polynom and
Field_theory. More importantly, performance should not decrease
significantly by adding new Morphism/Setoid declarations.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11337 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
| |
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11334 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
syntax yet. Doesn't change the auto/eauto behavior either.
- Typeclass resolution now considers everything transparent by
default and does it consistently for "open" and closed terms.
- Correctly declare singleton classes definition as opaque for proof
search.
- Add a few initial declarations to make iff, id, compose... opaque
- Add definition of dependent signatures for dependent function types
and remove corresponding exception code in class_tactics. The
instance requires higher-order unification and is not really usable yet.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11333 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
| |
intro-patterns and avoid useless generalizations on inductive
parameters.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11331 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now there is a class "SetoidRelation" for registering relations that
should always be considered as setoids and never unfolded. Every "Add
Relation" command adds an instance and impl,iff are there by
default. Now the test is: if there is a SetoidRelation instance, use it
; otherwise, allow unfolding to find an eq or fallback on
setoid_rewrite. To avoid searching for SetoidRelation instances
repeateadly we check that it is really needed first by unfolding the
hyp. Only two scripts relied on the now-forbidden semantics of rewriting
by an @eq inside a setoid relation, in Numbers.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11269 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
| |
destruction of records as a lot of scripts currently rely on it.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11263 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
| |
terms and [head_of_constr id] to put the head of a constr in some new
hypothesis. Both are used in a new tactic to deal with partial
applications in setoid_rewrite. Also fix bug #1905 by using
[subrelation] to take care of [Morphism (R ==> R') id] constraints.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11244 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
classes, and simplify the implementation.
- Experimental syntax {{ cl : Class args }} and (( cl : Class args ))
which respectively make cl an implicit or explicit argument ({{ }} is
equivalent to [ ]). Could be extended to any type of binder, eg.
[Definition flip ((R : relation carrier)) : relation carrier := ...].
The idea behind double brackets is to distinguish macro-binders which
perform implicit generalization from regular binders. It could also save
[ ] for other uses.
- Fix bug #1901 about {} binders in records.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11210 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Now [ id : Class foo ] makes id an explicit argument,
and [ Class foo ] is equivalent to [ {someid} : Class foo ].
This makes declarations such as "Class Ord [ eq : Eq a ]" have
sensible implicit args.
- Better handling of {} in class and record declarations, refactorize
code for declaring structures and classes.
- Fix merging of implicit arguments information on section closing.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11204 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
| |
- Cases on multiple objects
- Avoid dangerous coercion with evars in subtac_coercion
- Resolve typeclasses method-by-method to get better error messages.
- Correct merging of instance databases (and add debug printer)
- Fix a script in NOrder where a setoid_replace was not working before.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11198 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
goals containing existentials and use transparency information on
constants (optionally). Only used by the typeclasses eauto engine for
now, but could be used for other hint bases easily (just switch a boolean).
Had to add a new "creation" hint to be able to set said boolean upon
creation of the typeclass_instances hint db.
Improve the proof-search algorithm for Morphism, up to 10 seconds
gained in e.g. Field_theory, Ring_polynom. Added a morphism
declaration for [compose].
One needs to declare more constants as being unfoldable using
the [Typeclasses unfold] command so that discrimination is done correctly, but
that amounts to only 6 declarations in the standard library.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11184 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
| |
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11164 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
| |
letting the exception go uncatched.
- Reorder definitions in Morphisms, move the PER on (R ==> R') in
Morphisms where it can be declared properly. Add an instance for
Equivalence => Per.
- Change bug#1604 test file to the new semantics of [setoid_rewrite at]
- More unicode characters parsed by coqdoc.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11092 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Ajout clause "in" à "remember" (et passage du code en ML).
- Ajout clause "in" à "induction"/"destruct" qui, en ce cas, ajoute
aussi une égalité pour se souvenir du terme sur lequel l'induction
ou l'analyse de cas s'applique.
- Ajout "pose t as id" en standard (Matthieu: j'ai enlevé celui de
Programs qui avait la sémantique de "pose proof" tandis que le nouveau
a la même sémantique que "pose (id:=t)").
- Un peu de réorganisation, uniformisation de noms dans Arith, et
ajout EqNat dans Arith.
- Documentation tactiques et notations de tactiques.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11072 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
| |
Minor fix in Morphisms which prevented working with higher-order
morphisms and PER relations.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11065 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
change the default pretty-printing to use Π, λ instead of
forall and fun (and allow "," as well as "=>" for "fun" to be more
consistent with the standard forall and exists syntax). Parsing allows
theses new forms too, even if not in -unicode, and does not make Π or
λ keywords. As usual, criticism and suggestions are welcome :)
Not sure what to do about "->"/"→" ?
- [setoid_replace by] now uses tactic3() to get the right parsing level
for tactics.
- Type class [Instance] names are now mandatory.
- Document [rewrite at/by] and fix parsing of occs to support their
combination.
- Backtrack on [Enriching] modifier, now used exclusively in the
implementation of implicit arguments.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@10921 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
binders.
- Change syntax of type class instances to better match the usual syntax of
lemmas/definitions with name first, then arguments ":" instance.
Update theories/Classes accordingly.
- Correct globalization of tactic references when doing Ltac :=/::=, update
documentation.
- Remove the not so useful "(x &)" and "{{x}}" syntaxes from
Program.Utils, and subset_scope as well.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@10919 85f007b7-540e-0410-9357-904b9bb8a0f7
|