The new test ../../tests/lean/run/tactic27.lean demonstrates why we need this feature. The tactic 'apply @refl' is actually 'apply @refl.{?l}'. It is used inside of a repeat tactical. Each iteration of the 'repeat' may need to use a different value for ?l. Before this commit, there was not way to say we want a fresh ?l each iteration.
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
These two features make sense for solvers, but not in a general purpose tactic framework for building proofs like the one in Lean.
In most cases, we cannot build a counterexample anyway. These two features should be added in a custom framework for combining preprocessing techniques like in Z3.
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
This modification was motivated by a bug exposed by tst17 at tests/kernel/type_checker.
metavar_env is now a smart point to metavar_env_cell.
ro_metavar_env is a read-only smart pointer. It is useful to make sure we are using proof_state correctly.
example showing that the approach for caching metavar_env is broken in the type_checker
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
The environment object is a "smart-pointer".
Before this commit, the use of "const &" for environment objects was broken.
For example, suppose we have a function f that should not modify the input environment.
Before this commit, its signature would be
void f(environment const & env)
This is broken, f's implementation can easilty convert it to a read-write pointer by using
the copy constructor.
environment rw_env(env);
Now, f can use rw_env to update env.
To fix this issue, we now have ro_environment. It is a shared *const* pointer.
We can convert an environment into a ro_environment, but not the other way around.
ro_environment can also be seen as a form of documentation.
For example, now it is clear that type_inferer is not updating the environment, since its constructor takes a ro_environment.
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
After this commit, a value of type 'expr' cannot be a reference to nullptr.
This commit also fixes several bugs due to the use of 'null' expressions.
TODO: do the same for kernel objects, sexprs, etc.
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
