Draft
Conversation
Halbaroth
force-pushed
the
check-model-ci
branch
from
b66240a to
9aec94b
Compare
This PR fixes the issue OCamlPro#778. The suggested solution by this PR is as follows: - Collect user-declared symbols in `D_cnf`; - As the function `make` of the module `D_cnf` returns a list of commands, we add a new command `Decl` in `Frontend` to declare symbols; - The declared symbols are given to the SAT solver through a new function `declaration` in `sat_solver_sig`. - These symbols are collected in vectors and we keep indexes in another PR in order to manage properly the push/pop mechanism. - Finally the declared symbols are given to `Uf` and we create the empty model with them. Another annoying point comes from the current datastructure used in `ModelMap`. The type of `ModelMap` saves model values in the form of graphs. A graph is a finite set of constraints but there is no appropriate representation for the graph without constraints. Let's imagine we have the input: ```smt2 (set-option :produce-models true) (set-logic ALL) (declare-fun (Int) Int) (check-sat) (get-model) ``` We expect an output as follows: ```smt2 ( (define-fun f (_x Int) (@A as Int)) ) ``` But `Graph.empty` cannot hold the abstract value `@a`. A naive solution consists in using an ADT: ```ocaml type graph = | Empty of Id.typed | G of Graph.t ``` But we'll add an extra indirection. Probably the best solution would be to use a custom datastructure instead of a map to store the constraints. In this PR, we detect in `ModelMap.add` if the only constraint in the graph is of the form `something -> abstract value`. In this case we remove this constraint. This solution isn't perfect as explained in issue OCamlPro#1018. The biggest drawback of this solution is the fact we have to declare symbols in the SAT API, otherwise models returned by `get_model` could be incomplete.
Halbaroth
force-pushed
the
check-model-ci
branch
from
9aec94b to
f91e02c
Compare
Collaborator
Author
|
As I explained during the last dev meeting, we cannot check models of problems involving custom builtin functions like But I believe we can do it by adding definitions in the input problems. I think this isn't very important and I'll finish this PR without |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Please do not merge this PR on
nextbefore #789.