Commit graph

167 commits

Author SHA1 Message Date
Leonardo de Moura
98897b467d feat(frontends/lean/parser): add support for Lua expression code blocks
In expression code blocks, we do not have to write a "return".
After this commit, the argument of an apply command is a Lua expression instead of a Lua block of code. That is, we can now write

apply (** REPEAT(ORELSE(imp_tactic, conj_tactic, conj_hyp_tactic, assumption_tactic)) **)

instead of

apply (** return REPEAT(ORELSE(imp_tactic, conj_tactic, conj_hyp_tactic, assumption_tactic)) **)

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-29 10:21:24 -08:00
Leonardo de Moura
f9874cd675 feat(library/tactic): add to_tactic_ext, it allows functions that return tactics to be used where a tactic is expected
For example, after this commit, we can write

simple_tac = REPEAT(ORELSE(imp_tactic, conj_tactic)) .. assumption_tactic

instead of

simple_tac = REPEAT(ORELSE(imp_tactic(), conj_tactic())) .. assumption_tactic()

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-29 09:40:21 -08:00
Leonardo de Moura
066dacea31 feat(library/tactic): add show_tactic, and optional '.' in the end of tactic command
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-29 01:33:26 -08:00
Leonardo de Moura
6832b265e9 style(frontends/lean/parse): missing include
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-28 21:50:55 -08:00
Leonardo de Moura
5dfb3b8b56 feat(frontends/lean/parse): allow script-code blocks to be used in the apply command
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-28 21:48:30 -08:00
Leonardo de Moura
dae86c2ffa feat(frontends/lean/parser): add basic tactic support in the frontend
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-28 21:08:12 -08:00
Leonardo de Moura
a2aa90ae66 refactor(util/script_state): replace std::recursive_mutex with std::mutex, and use unlock_guard
The unlock_guard and exec_unprotected will be useful also for implementing the Lua tactic API.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-27 15:40:41 -08:00
Leonardo de Moura
f7e8545e97 refactor(frontends/lua): rename leanlua_state to script_state, and move it to util
This commit also minimizes the dependencies of script_state.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-27 14:57:36 -08:00
Leonardo de Moura
d87ad9eb7e refactor(util/lua): propagate C++ Lean exceptions in Lua
The following call sequence is possible:
C++ -> Lua -> C++ -> Lua -> C++

The first block of C++ is the Lean main function.
The main function invokes the Lua interpreter.
The Lua interpreter invokes a C++ Lean API.
Then the Lean API invokes a callback implemented in Lua.
The Lua callback invokes another Lean API.
Now, suppose the Lean API throws an exception.
We want the C++ exception to propagate over the mixed C++/Lua call stack.
We use the clone/rethrow exception idiom to achieve this goal.

Before this commit, the C++ exceptions were converted into strings
using the method what(), and then they were propagated over the Lua
stack using lua_error. A lua_error was then converted into a lua_exception when going back to C++.
This solution was very unsatisfactory, since all C++ exceptions were being converted into a lua_exception, and consequently the structure of the exception was being lost.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-27 12:25:29 -08:00
Leonardo de Moura
5346b67651 refactor(library/state): rename Lean state object to io_state
The idea is to make it clear that io_state is distinguish it from proof_state, and from leanlua_state.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-21 17:29:06 -08:00
Leonardo de Moura
0126fa0499 refactor(kernel): add find_fn, replace for_each_fn with find_fn when appropriate, remove unnecessary function has_cached_type
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-19 13:03:46 -08:00
Leonardo de Moura
5cfcb7e144 chore(kernel/for_each): use consistent naming convetions
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-19 11:24:02 -08:00
Leonardo de Moura
69be5f6c94 feat(kernel/environment): track which modules were already imported
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-17 18:15:44 -08:00
Leonardo de Moura
8525e8534b feat(lua): expose parse_expr and parse_commands from frontends/lean in the Lua API
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-15 16:11:26 -08:00
Leonardo de Moura
19533c811b feat(library/script_evaluator): add abstract class that exposes only the API needed by frontend objects
The main motivation is to break the remove the dependency frontends/lean <-- bindings/lua.
This dependency is undesirable because we want to expose the frontends/lean parser and pretty printer objects at bindings/lua.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-15 12:13:09 -08:00
Leonardo de Moura
691893258d feat(kernel/expr): add hash code based on allocation time
The new hash code has the property that given expr_cell * c1 and expr_cell * c2,
if c1 != c2 then there is a high propbability that c1->hash_alloc() != c2->hash_alloc().

The structural hash code hash() does not have this property because we may have
c1 != c2, but c1 and c2 are structurally equal.

The new hash code is only compatible with pointer equality.
By compatible we mean, if c1 == c2, then c1->hash_alloc() == c2->hash_alloc().
This property is obvious because hash_alloc() does not have side-effects.

The test tests/lua/big.lua exposes the problem fixed by this commit.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-14 02:43:11 -08:00
Leonardo de Moura
eacd60de9c feat(frontends/lean): return the operator associated with constant expressions that are names of builtin values
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-13 16:08:21 -08:00
Leonardo de Moura
c4c548dc5d feat(*): simplify interrupt propagation
Instead of having m_interrupted flags in several components. We use a thread_local global variable.
The new approach is much simpler to get right since there is no risk of "forgetting" to propagate
the set_interrupt method to sub-components.

The plan is to support set_interrupt methods and m_interrupted flags only in tactic objects.
We need to support them in tactics and tacticals because we want to implement combinators/tacticals such as (try_for T M) that fails if tactic T does not finish in M ms.
For example, consider the tactic:

    try-for (T1 ORELSE T2) 5

It tries the tactic (T1 ORELSE T2) for 5ms.
Thus, if T1 does not finish after 5ms an interrupt request is sent, and T1 is interrupted.
Now, if you do not have a m_interrupted flag marking each tactic, the ORELSE combinator will try T2.
The set_interrupt method for ORELSE tactical should turn on the m_interrupted flag.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-12 21:45:48 -08:00
Leonardo de Moura
9a5f86fce6 feat(lua): use (** ... **) instead of {{ ... }} for nested Lua scripts
The token }} is a bad delimiter for blocks of Lua script code nested in Lean files.
The problem is that the sequence }} occurs very often in Lua code because Lua uses { and } to build tables/lists/arrays.
Here is an example of Lua code that contains the sequence }}
     t = {{1, 2}, {2, 3}, {3, 4}}

In Lean, (* ... *) is used to create comments. Thus, (** ... **) code blocks will not affect
valid Lean files. It also looks reasonably good.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-12 16:05:49 -08:00
Leonardo de Moura
8190d4fed5 feat(lua): allow Lua scripts to update 'global' options
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-12 15:38:00 -08:00
Leonardo de Moura
31abc00db8 chore(*): add LCOV_EXCL_LINE to lean_unreachable statements
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-11 09:19:38 -08:00
Leonardo de Moura
7683188ab0 chore(emplace_back): use emplace_back when appropriate
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-10 11:14:04 -08:00
Leonardo de Moura
8c140ff86f feat(lua): allow lua scripts (embedded in Lean files) to access the environment
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-10 11:14:04 -08:00
Leonardo de Moura
57b9657bf0 feat(lua): add lua_exception for wrapping lua errors, and improve Lua error messages in the Lean frontend
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-07 15:19:26 -08:00
Leonardo de Moura
a9b2be0b9c feat(frontends/lean): add support for embedded Lua scripts in Lean files
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-07 13:56:04 -08:00
Leonardo de Moura
8012c4f644 fix(kernel/environment): add weak reference to environment objects
We need weak references to environment objects because the environment has a reference to the type_checker and the type_checker has a reference back to the environment. Before, we were breaking the cycle using an "environment const &". This was a dangerous hack because the environment smart pointer passed to the type_checker could be on the stack. The weak_ref is much safer.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-07 11:29:08 -08:00
Leonardo de Moura
9c60eed93c refactor(kernel/metavar): avoid using unique names for default metavariable prefix
The problem is that unique names depend on the order compilation units are initialized. The order of initialization is not specified by the C++ standard. Then, different compilers (or even the same compiler) may produce different initialization orders, and consequently the metavariable prefix is going to be different for different builds. This is not a bug, but it makes unit tests to fail since the output produced by different builds is different for the same input file.
Avoiding unique name feature in the default metavariable prefix avoids this problem.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-07 10:16:25 -08:00
Leonardo de Moura
2141ee12f4 refactor(frontends/lean): use extension objects to store lean default frontend data in the environment
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-07 10:00:12 -08:00
Soonho Kong
044813615e fix: add '#include <tuple>' 2013-11-03 13:00:42 -05:00
Leonardo de Moura
bf998d8661 feat(frontends/lean/parser): allow 'typeless' definitions, the type is inferred by the system
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-01 08:51:49 -07:00
Leonardo de Moura
fdea8aba10 feat(frontends/lean/scanner): allow '#' to be used in class B identifiers
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-11-01 08:42:25 -07:00
Leonardo de Moura
96dcd003c6 fix(frontends/lean/parser): associated position with 'type' placeholder
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-31 16:27:36 -07:00
Leonardo de Moura
aa99ac6618 feat(kernel/value): allow semantic attachments to use coercions when being pretty printed
For example, this feature is useful when displaying the integer value 10 with coercions enabled. In this case, we want to display "nat_to_int 10" instead of "10".

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-30 11:42:26 -07:00
Leonardo de Moura
032f5cd7b3 feat(frontends/lean): make the 'expression template' argument in Subst implicit because higher-order matching can infer it.
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-30 10:45:43 -07:00
Leonardo de Moura
bc92671ae4 fix(frontends/lean/notation): adjust the implicit arguments of TransExt, and add new test
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-29 17:12:50 -07:00
Leonardo de Moura
4dd6cead83 refactor(equality): make homogeneous equality the default equality
It was not a good idea to use heterogeneous equality as the default equality in Lean.
It creates the following problems.

- Heterogeneous equality does not propagate constraints in the elaborator.
For example, suppose that l has type (List Int), then the expression
     l = nil
will not propagate the type (List Int) to nil.

- It is easy to write false. For example, suppose x has type Real, and the user
writes x = 0. This is equivalent to false, since 0 has type Nat. The elaborator cannot introduce
the coercion since x = 0 is a type correct expression.

Homogeneous equality does not suffer from the problems above.
We keep heterogeneous equality because it is useful for generating proof terms.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-29 16:20:06 -07:00
Leonardo de Moura
d0009d0242 feat(frontends/lean): make the first argument of if-expression implicit, add support for marking implicit arguments on builtin symbols (aka semantic attachments)
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-29 15:53:50 -07:00
Leonardo de Moura
2d88922543 feat(frontends/lean/elaborator): solve easy overloads at preprocessing time
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-29 10:07:15 -07:00
Leonardo de Moura
4564bfa1d3 feat(library/elaborator): improve simple_ho_match
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-27 11:17:03 -07:00
Leonardo de Moura
dbefc91151 fix(kernel/metavar): add normalize assignment justification
We need that when we normalize the assignment in a metavariable environment.
That is, we replace metavariable in a substitution with other assignments.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-27 11:02:34 -07:00
Leonardo de Moura
92f5a31976 feat(kernel/expr): add new mk_app template for creating applications using a collection
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-27 08:53:58 -07:00
Leonardo de Moura
57d9d23bd4 feat(kernel/for_each): allow function F to interrupt for_each search
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-25 14:58:02 -07:00
Leonardo de Moura
e765105ea5 fix(frontends/lean/pp): let expressions
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-25 12:30:39 -07:00
Leonardo de Moura
f0e149d77b fix(frontends/lean/pp): fix how Type expressions are pretty printed
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 20:02:34 -07:00
Leonardo de Moura
65a514ad8a fix(frontends/lean/parser): option propagation
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 19:04:06 -07:00
Leonardo de Moura
576b4e2169 fix(frontends/lean/pp): missing comma when printing contexts
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 19:00:24 -07:00
Leonardo de Moura
df07a84d11 feat(frontends/lean/parser): display elaborator error messages
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 18:48:53 -07:00
Leonardo de Moura
e3efe39eeb fix(elaborator): fix bug in higher-order matching/unification
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 17:53:37 -07:00
Leonardo de Moura
d1a2a4ea7e feat(frontends/lean/elaborator): add support for coercions in let declarations
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 17:01:06 -07:00
Leonardo de Moura
872b698bc3 feat(elaborator): add option m_assume_injectivity for getting more concise solutions
We may miss solutions, but the solutions found are much more readable.
For example, without this option, for elaboration problem

Theorem Example4 (a b c d e : N) (H: (a = b ∧ b = e ∧ b = c) ∨ (a = d ∧ d = c)) : (h a c) = (h c a) :=
    DisjCases H
              (fun H1 : _,
                   let AeqC := Trans (Conjunct1 H1) (Conjunct2 (Conjunct2 H1))
                   in CongrH AeqC (Symm AeqC))
              (fun H1 : _,
                   let AeqC := Trans (Conjunct1 H1) (Conjunct2 H1)
                   in CongrH AeqC (Symm AeqC))

the elaborator generates

Theorem Example4 (a b c d e : N) (H : a = b ∧ b = e ∧ b = c ∨ a = d ∧ d = c) : (h a c) = (h c a) :=
    DisjCases
        H
        (λ H1 : if
               Bool
               (if Bool (a = b) (if Bool (if Bool (if Bool (b = e) (if Bool (b = c) ⊥ ⊤) ⊤) ⊥ ⊤) ⊥ ⊤) ⊤)
               ⊥
               ⊤,
           let AeqC := Trans (Conjunct1 H1) (Conjunct2 (Conjunct2 H1)) in CongrH AeqC (Symm AeqC))
        (λ H1 : if Bool (if Bool (a = d) (if Bool (d = c) ⊥ ⊤) ⊤) ⊥ ⊤,
           let AeqC := Trans (Conjunct1 H1) (Conjunct2 H1) in CongrH AeqC (Symm AeqC))

The solution is correct, but it is not very readable. The problem is that the elaborator expands the definitions of \/ and /\.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2013-10-24 16:47:50 -07:00