When instantiating a term, we had an attempt at an optimisation that
assumed variables would be in scope order. But it's not safe to make
that assumption, and it turns out it makes no measurable difference to
performance anyway.
Instead of using a raw name, which is error prone, use a Var so we have
a proper reference into the environment. This is important when we're
combining case blocks and where clauses, because we need to rebuild the
environment properly for calls to functions defined in the where block.
The old way only worked by chance, because the argumemt order happens to
be the same in all cases. I noticed due to some experiments elsewhere
with different ways of elaborating case, which broke that assumption.
The meaning of the list of Vars is actually the opposite of what it was
taken to be... fortunately, the performance works out roughly the same.
Also this way is (arguably) simpler, which is usually a good sign.
This is quite fiddly as it the blocks might be in different contexts so
we need to keep track of which variables correspond in the scrutinees of
the blocks. Once that's done, check the terms at the leaves convert,
then check the corresponding variables convert.
This may not be perfect yet, because we only look at case scrutinees to
find correspondence. It might also be a bit slower than it could be, but
at least these checks are quite rare.
Fixes#208 and maybe some others?
Allows quoting a term back to a TTImp. Test reflection007 shows one
possible use for this, building a reflected, type safe, representation
of an expression.
So the type of Elab now gives the expected type that's being elaborated
to, meaning that we can run 'check' in the middle of scripts and use the
result.
On the LHS, we want to match against the reflected thing, so FC and
implicits need to turn into match anything patterns, or we won't match
anything at all. This means we can put quoted terms on the LHS, with
pattern variables under ~().
By default, the search depth for auto implcit arguments is limited to
50. This patch adds a new pragma `%auto_implicit_depth` with which the
user can change the depth limit.
Instead of merely generating a locally unique name, use the existing
code for generating a new unique variable name in the unifier, which is
therefore globally unique.
Get the names of local variables. and add the ability to look up their
types.
When we get a reflected TTImp, either checking the Goal or looking up a
type, it's not impossible that there'll be some repeated binder names,
so also make sure binders are unique relative to the current context.
Ideally we'd also rename things in the environment to guarantee that all
names are unique, but we don't yet.
(This would be much easier if reflected terms were typed such that they
were well scoped, but that would also make reflection harder to use.)
This invokes a script of type Elab (). %runElan in a term invokes a
script of type Elab TT. The elaborator now pushes in that type, so that
it'll report an appropriate error if you give it a script of the wrong
type.
A %macro must always be fully applied. Whenever the elaborator
encounters a %macro application (except in a function LHS) it evaluates
the application and sends the result to %runElab. So:
%macro
foo : args -> Elab TT
...
def = foo a b c
is equivalent to
foo : args -> Elab TT
...
def = %runElab foo a b c
