| Age | Commit message (Collapse) | Author |
|---|
|
Also changed the API of pr_subgoals now using labels.
Also removed a trailing space in printing existentials.
|
|
There is more churn than there should be because SearchPattern uses a
non-local sorting algorithm; the comparison function considers many
constants equal in priority and leaves it up to the heap structure to
break ties, which seems wrong. This has been reported as
[bug #5573](https://coq.inria.fr/bugs/show_bug.cgi?id=5573).
|
|
Concretely, we bypass the following limitation: The notation
"{ ' pat | P }" broke the parsing of expressions of the form
"{ forall x, P } + { Q }". Indeed the latter works thanks to a
tolerance of Camlp5 in parsing "forall x, P" at level 200 while the
rule asks to actually parse the interior of "{ ... }" at level 99 (the
reason for 99 is to be below the rule for "M : T" which is at level
100, so that "{ x : A | P }" does not see "x : A" as a cast). Adding
an extra "'"; pat = pattern in parallel to c = constr LEVEL "99" broke
the tolerance for constr at level 200.
We fix this by adding an ad hoc rule for "{ binder_constr }" in the
native grammar (g_constr.ml4).
Actually, this is inconsistent with having a rule for "{ constr at level 99 }"
in Notations.v. We should have both rules in Notations.v or both rules
hard-wired in the native grammar. But I still don't know what is the best
decision to take, so leaving as it at the current time.
Advantages of hard-wiring both rules in g_constr.ml4: a bit simpler in
metasyntax.ml (no need to ensure that the rule exist). Disadvantages:
if one wants a different initial state without the business needing
the "{ }" for sumbool, sumor, sig, sigT, one still have the rules
there.
Advantages of having them in Notations.v: more modular, we can change
the initial state. Disadvantages: one would need a new kind of
modifier, something like "x at level 99 || binder_constr", with all
the difficulty to find a nice, intuitive, name for "binder_constr",
and the difficulty of understanding if there is a generality to this
"||" disjunction operator, and whether it should be documented or not.
|
|
|
|
For compatibility, the default is to parse as ident and not as pattern.
|
|
Concretely, we provide "constr as ident", "constr as strict pattern"
and "constr as pattern".
This tells to parse a binder as a constr, restricting to only ident or
to only a strict pattern, or to a pattern which can also be an ident.
The "strict pattern" modifier allows to restrict the use of patterns
in printing rules. This allows e.g. to select the appropriate rule for
printing between {x|P} and {'pat|P}.
|
|
We reason up to order, and accept to match a final catch-all clauses
with any other clause.
This allows for instance to parse and print a notation of the form
"if t is S n then p else q".
|
|
This now works not only for parsing of fun/forall (as in 8.6), but
also for arbitraty notations with binders and for printing.
|
|
Example which is now reprinted as parsed:
fun '((x,y) as z) => (y,x)=z
|
|
To deal with existing notations starting with a "let" (see notation
"for" in output/Notation2.v) we adopt the pragmatic approach of
glueing only inner let by default.
Ideally, it might be nicer to detect if there is an overlap of
notation, and not to glue only in case of overlap. We could also
decide that a notation starting with a "let" should always be
protected by some constant, say "id", so as to avoid possible
collisions, but this would require changes user side.
|
|
For historical reasons (this was one of the first examples of
notations with binders), there was a special treatment for notations
whose right-hand side had the form "forall x, P" or "fun x => P". Not
only this is not necessary, but this prevents notations binding to
expressions such as "forall x, x>0 -> P" to be used in printing.
We let the general case absorb this particular case.
We add the integration of "let x:=c in ..." in the middle of a
notation with recursive binders as part of the binder list, reprinting
it "(x:=c)" (this was formerly the case only for the above particular
case).
Note that integrating "let" in sequence of binders is stil not the
case for the regular "forall"/"fun". Should we?
|
|
|
|
This allows in particular to define notations with 'pat style binders.
E.g.:
A non-trivial change in this commit is storing binders and patterns
separately from terms.
This is not strictly necessary but has some advantages.
However, it is relatively common to have binders also used as terms,
or binders parsed as terms. Thus, it is already relatively common to
embed binders into terms (see e.g. notation for ETA in output test
Notations3.v) or to coerce terms to idents (see e.g. the notation for
{x|P} where x is parsed as a constr).
So, it is as simple to always store idents (and eventually patterns)
as terms.
|
|
For instance, the following is now possible:
Check {(x,y)|x+y=0}.
Some questions remains. Maybe, by consistency, the notation should be
"{'(x,y)|x+y=0}"...
|
|
|
|
Seizing this opportunity to generalize the possibility for different
associativity into simply reversing the order or not. Also dropping
some dead code.
Example of recursive notation now working:
Notation "[ a , .. , b |- A ]" := (cons b .. (cons a nil) .., A).
|
|
This allows for instance to support recursive notations of the form:
Notation "! x .. y # A #" :=
(((forall x, x=x),(forall x, x=0)), .. (((forall y, y=y),(forall y, y=0)), A) ..)
(at level 200, x binder).
|
|
This makes treatment of recursive binders closer to the one of
recursive terms. It is unclear whether there are interesting notations
liable to use this, but this shall make easier mixing recursive
binders and recursive terms as in next commits.
Example of (artificial) notation that this commit supports:
Notation "! x .. y # A #" :=
(.. (A,(forall x, True)) ..,(forall y, True))
(at level 200, x binder).
|
|
- The "terminator" of a recursive notation is now interpreted in the
environment in which it occurs rather than the environment at the
beginning of the recursive patterns.
Note that due to a tolerance in checking unbound variables of
notations, a variable unbound in the environment was still working
ok as long as no user-given variable was shadowing a private
variable of the notation - see the "exists_mixed" example in
test-suite.
Conversely, in a notation such as:
Notation "!! x .. y # A #" :=
((forall x, True), .. ((forall y, True), A) ..)
(at level 200, x binder).
Check !! a b # a=b #.
The unbound "a" was detected only at pretyping and not as expected
at internalizing time, due to "a=b" interpreted in context
containing a and b.
- Similarly, each binder is now interpreted in the environment in
which it occurs rather than as if the sequence of binders was
dependent from the left to the right (such a dependency was ok for
"forall" or "exists" but not in general).
For instance, in:
Notation "!! x .. y # A #" :=
((forall x, True), .. ((forall y, True), A) ..)
(at level 200, x binder).
Check !! (a:nat) (b:a=a) # True #.
The illegal dependency of the type of b in a was detected only at
pretyping time.
- If a let-in occurs in the sequence of binders of a notation with a
recursive pattern, it is now inserted in between the occurrences of
the iterator rather than glued with the forall/fun of the iterator.
For instance, in:
Notation "'exists_true' x .. y , P" :=
(exists x, True /\ .. (exists y, True /\ P) ..)
(at level 200, x binder).
Check exists_true '(x,y) (u:=0), x=y.
We now get
exists '(x, y), True /\ (let u := 0 in True /\ x = y)
while we had before the let-in breaking the repeated pattern:
exists '(x, y), (let u := 0 in True /\ x = y)
This is more compositional, and, in particular, the printer algorithm
now recognizes the pattern which is otherwise broken.
|
|
This works for contexts in Definition and co, but not yet for "fun" and co.
|
|
The motivations are:
- To reflect the concrete syntax more closely.
- To factorize the different places where "contexts" are internalized:
before this patch, there is a different treatment of `Definition f
'(x,y) := x+y` and `Definition f := fun '(x,y) => x+y`, and a hack
to interpret `Definition f `pat := c : t`. With the patch, the fix
to avoid seeing a variable named `pat` works for both `fun 'x =>
...` and `Definition f 'x := ...`.
The drawbacks are:
- Counterpart to reflecting the concrete syntax more closerly, there
are more redundancies in the syntax. For instance, the case `CLetIn
(na,b,t,c)` can appears also in the form `CProdN (CLocalDef
(na,b,t)::rest,d)` and `CLambdaN (CLocalDef (na,b,t)::rest,d)`.
- Changes in the API, hence adaptation of plugins referring to `constr_expr` needed.
|
|
Was apparently forgotten in a67bd7f9.
|
|
|
|
|
|
|
|
There is no way today to distinguish primitive projections from
compatibility constants, at least in the case of a record without
parameters.
We remedy to this by always using the r.(p) syntax when printing
primitive projections, even with Set Printing All.
The input syntax r.(p) is still elaborated to GApp, so that we can preserve
the compatibility layer. Hopefully we can make up a plan to get rid of that
layer, but it will require fixing a few problems.
|
|
|
|
|
|
This closes #5082 and closes #5778, but makes #6404 apply to `abstract`
as well as `transparent_abstract`. This is unfortunate, but I think it
is worth it to get `abstract` in the profile (and therefore not
misassign the time spent sending the subproof to the kernel). Another
alternative would have been to add a dedicated entry to `ltac_call_kind`
for `TacAbstract`, but I think it's better to just deal with #6404 all
at once.
The "better" solution here would have been to move `abstract` out of the
Ltac syntax tree and define it via `TACTIC EXTEND` like
`transparent_abstract`. However, the STM relies on its ability to
recognize `abstract` and `solve [ abstract ... ]` syntactically so that
it can handle `par: abstract`.
Note that had to add locations to `TacAbstract` nodes, as I could not
figure out how to call `push_trace` otherwise.
|
|
|
|
Adding a "let-in"-sensitive function hnf_prod_applist_assum to
instantiate parameters and using it for printing.
Thanks to PMP for reporting.
|
|
Moreover, when there are at least two clauses and the last most
factorizable one is a disjunction with no variables, turn it into a
catch-all clause.
Adding options
Unset Printing Allow Default Clause.
to deactivate the second behavior, and
Unset Printing Factorizable Match Patterns.
to deactivate the first behavior (deactivating the first one
deactivates also the second one).
E.g. printing
match x with Eq => 1 | _ => 0 end
gives
match x with
| Eq => 1
| _ => 0
end
or (with default clause deactivates):
match x with
| Eq => 1
| Lt | Gt => 0
end
More to be done, e.g. reconstructing multiple patterns in Nat.eqb...
|
|
Fixes GH#6384 and GH#6385.
|
|
|
|
to use among several of them
|
|
|
|
|
|
|
|
|
|
|
|
- Regularly declared for for polymorphic constants
- Declared globally for monomorphic constants.
E.g mono@{i} := Type@{i} is printed as
mono@{mono.i} := Type@{mono.i}.
There can be a name clash if there's a module and a constant of the
same name. It is detected and is an error if the constant is first
but is not detected and the name for the constant not
registered (??) if the constant comes second.
Accept VarRef when registering universe binders
Fix two problems found by Gaëtan where binders were not registered properly
Simplify API substantially by not passing around a substructure of an
already carrier-around structure in interpretation/declaration code of
constants and proofs
Fix an issue of the stronger restrict universe context + no evd leak
This is uncovered by not having an evd leak in interp_definition, and
the stronger restrict_universe_context. This patch could be backported
to 8.7, it could also be triggered by the previous restrict_context I
think.
|
|
|
|
This allows us to enforce that it works without breaking the build
when it doesn't.
|
|
See discussion on coq-club starting on 23 August 2016.
An open question: what priority to give to "abbreviations"?
|
|
|
|
Note that this makes the following syntax valid:
Axiom foo@{i} bar : Type@{i}.
(ie putting a universe declaration on the first axiom in the list, the
declaration then holds for the whole list).
|
|
Also nicer error when the constraints are impossible.
|
|
|
|
|
|
|
