WIP: CEG-Prophecy implementation in Pono #69
Conversation
…s by always using relational system
… Abstractor a friend
engines/ceg_prophecy_arrays.h
Outdated
| ProphecyModifier pm_; | ||
|
|
||
| size_t num_added_axioms_ = | ||
| 0; ///< set by refine to the number of added axioms |
There was a problem hiding this comment.
do the assignment in the initialize method
modifiers/array_abstractor.cpp
Outdated
| abs_rts.set_init(abs_init); | ||
| abs_rts.set_trans(abs_trans); | ||
|
|
||
| // TODO: remove these debug prints |
There was a problem hiding this comment.
should we remove it now or do you want to do it later in a separate PR?
There was a problem hiding this comment.
Let's just do it now, thanks!
pono.cpp
Outdated
| logger.set_verbosity(pono_options.verbosity_); | ||
|
|
||
| try { | ||
| // TEMP comment this out for development |
There was a problem hiding this comment.
anything to be done here?
There was a problem hiding this comment.
Yep, thanks. It does make sense to catch any exceptions, right? Or do you think it's better to let them go uncaught?
engines/ceg_prophecy_arrays.cpp
Outdated
| return ProverResult::FALSE; | ||
| } | ||
| reached_k_++; | ||
| } while (num_added_axioms_); |
There was a problem hiding this comment.
when will it exit from the loop? (only in the case of !refine())
There was a problem hiding this comment.
Either for !refine() or when no axioms were added (e.g. num_added_axioms_ == 0). So it tries to refine bmc until no new axioms were found, or there's a concrete counterexample.
engines/ceg_prophecy_arrays.cpp
Outdated
| ProverResult CegProphecyArrays::prove() | ||
| { | ||
| ProverResult res = ProverResult::FALSE; | ||
| while (res == ProverResult::FALSE) { |
There was a problem hiding this comment.
why are we refining before checking the abstract property?
There was a problem hiding this comment.
Ahh it's just a heuristic -- we used that in the prototype as well. It adds a few axioms first before trying to prove it. If the property doesn't require any axioms, then it will refine quickly anyway and get to the underlying model checker.
engines/ceg_prophecy_arrays.cpp
Outdated
| } | ||
|
|
||
| ProverResult CegProphecyArrays::prove() | ||
| { |
There was a problem hiding this comment.
also, maybe use check_until with INT_MAX
There was a problem hiding this comment.
Makes sense! That means it will call check_until(INT_MAX) for the underlying model checker instead of prove. So far that means exactly the same thing, but I wasn't sure it would stay that way. I'll make the change, but just thought I'd point that out.
|
Thanks for the review @ahmed-irfan! I know it was a lot. I addressed some your comments in the latest commit and commented about the rest. Let me know your thoughts when you get a chance. |
|
@makaimann need to resolve conflicts |
No description provided.