Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
Fixes #3353. #3062 introduced a flag that tracks whether we have seen a
`(set-logic ...)` command to improve the handling of `--force-logic`.
However, the flag was not set to `false` when `(reset)` was called. This
commit fixes the issue.
|
|
|
|
* Towards fix for non-linear models
* Format
* Fix
* More
* Improve
* Format
* More
|
|
|
|
|
|
PR #3388 didn't disable the regression correctly (due to using `REQUIRE`
instead of `REQUIRES`). This commit fixes the issue.
|
|
This commit disables a regression test that was failing for the
competition build due to not emitting the expected error message.
|
|
Fixes #1399.
|
|
|
|
CVC4 was printing success when `--force-logic` was used because
internally, `--force-logic` generates a `SetBenchmarkLogicCommand`. This
caused issues with the SMT-COMP trace executor. This commit fixes the
behavior by muting the command if it was not issued by the user.
The issue was likely introduced with #3062.
|
|
This makes `--sygus-inference` a no-op for inputs where there is a free function whose sort cannot be handled in a sygus grammar.
It also fixes an issue where skolem variables were not being treated as functions-to-synthesize.
Fixes #3250 and fixes #3356.
|
|
|
|
|
|
The ackermannization process is currently already support general theories rather than specifically for BV. In this pull request, an option has been added to turn on ackermannization independently.
|
|
This commit makes CVC4 complain if the user asked for proofs for an unsupported logic (in this contest, ALL is considered unsupported).
Changes in the regression script are introduced as well, in order to only request proofs for regressions in supported logics.
|
|
This short circuits CEGQI when the conjecture is solvable by simple equality reasoning. It adds two examples where we previously would have fallen back on enumeration due to not having an instantiation technique for strings, despite the conjectures being trivially solvable.
|
|
|
|
|
|
|
|
constructors (#3259)
|
|
|
|
This commit removes the SMT1 parser infrastructure and adds the SMT2 translations of the SMT1 regression tests. For now this commit removes regression test regress3/pp-regfile.smt since the SMT2 translation has a file size of 887M (vs. 172K for the SMT1 version).
Fixes #2948 and fixes #1313.
|
|
|
|
To make sure that our `str.code` function is injectve (except for -1 in
the codomain), we send the inference that `str.code(x) == -1 v
str.code(x) != str.code(y) v x == y` for each pair of `str.code` terms.
Because of the order of disjuncts, `str.code(x) != str.code(y)` was
usually assigned true. This in turn lead to a difficult problem for the
arithmetic engine if there were more `str.code` applications than the
size of the domain. E.g. if we had `0 <= str.code(xi) < 10` for 0 <= i
<= 10, then the arithmetic engine had a difficult time finding a
conflict. This PR improves the heuristic by setting the phase of
`str.code(x) != str.code(y)` to false, so we prefer to keep the
`str.code` values equal instead of trying to make them different.
This change is also reflected in the models produced for inputs
involving `str.code`: Previously, we were producing models with
different values for the `str.code` whereas now the models are much more
uniform.
The PR adds two regressions, one testing `str.code` performance directly
and one testing it for `str.code` terms generated by `re.range`.
Signed-off-by: Andres Noetzli <anoetzli@amazon.com>
|
|
We have a conflict if we have `str.in.re(x, R1)` and `~str.in.re(x, R2)`
and `R2` includes `R1` because there is no possible value for `x` that
satisfies both memberships. This commit adds code to detect regular
expression inclusion for a small fragment of regular expressions: string
literals with single char (`re.allchar`) and multichar wildcards
(`re.*(re.allchar)`).
Signed-off-by: Andres Noetzli <anoetzli@amazon.com>
|
|
|
|
|
|
The regression script did not extract the expected status from incremental
SMT-LIB v2 benchmarks correctly if status was given via (set-info :status ...). The
script used re.search for finding the status, which only searches for
the first occurrence instead of finding all (set-info :status ...).
This commit fixes the issue by using re.findall instead.
|
|
* Removes incremental API check (#3011)
* Fixes toSatValueLit to use the new semantics of CaDiCaL's val()
Fixes #3011
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The `--force-logic` command line argument can be used to override a
logic specified in an input file or to set a logic when none is given.
Before this commit, both the `SmtEngine` and the parser were aware of
that argument. However, there were two issues if an input file didn't
specify a logic but `--force-logic` was used:
- Upon parsing `--force-logic`, the `SmtEngine` was informed about it
and set the logic to the forced logic. Then, the parser detected that
there was no `set-logic` command, so it set the logic to `ALL` and
emitted a corresponding warning. Finally, `SmtEngine::setDefaults()`
detected that `forceLogic` was set by the user and changed the logic
back to the forced logic. The warning was confusing and setting the
logic multiple times was not elegant.
- For eager bit-blasting, the logic was checked before resetting the
logic to the forced logic, so it would emit an error that eager
bit-blasting couldn't be used with the logic (which was `ALL` at that
point of the execution). This was a problem in the competition because
our runscript parses the `set-logic` command to decide on the
appropriate arguments to use and passes the logic to CVC4 via
`--force-logic`.
This commit moves the handling of `--force-logic` entirely into the
parser. The rationale for that is that this is not an API-level issue
(if you use the API you simply set the logic you want, forcing a
different logic in addition doesn't make sense) and simplifies the
handling of the option (no listeners need to be installed and the logic
is set only once). This commit also removes the option to set the logic
via `(set-option :cvc4-logic ...)` because it complicates matters (e.g.
which method of setting the logic takes precedence?). For the CVC and
the TPTP languages the commit creates a command to set the logic in
`SmtEngine` when the logic is forced in the parser instead of relying on
`SmtEngine` to figure it out itself.
|
|
|
|
`let_shadowing.smt2` uses dumping to test our printing infrastructure.
Since some builds do not support dumping, this commit disables that
regression for non-dumping builds. Additionally, it enables an error
message when trying to dump with a muzzled build and corrects the output
of `--show-config` to indicate that muzzled builds cannot dump.
Previously, the dumping output of a muzzled build was just silently
empty.
Most of the changes in `dump.cpp` are due to reformatting.
|
|
Fixes #3005. When printing nodes, we introduce `let` expressions on the
fly. However, when doing that, we have to be careful that we don't
shadow existing variables with the same name. When quantifiers are
involved, we do not descend into the quantifiers to avoid letifying
terms with bound variables that then go out of scope (see #1863). Thus,
to avoid shadowing variables appearing in quantifiers, we have to
collect all the variables appearing in that term to make sure that the
let does not shadow them. In #3005, the issue was caused by a `let` that
was introduced outside of a quantifier and then was shadowed in the body
of the quantifier by another `let` introduced for that body.
|
|
Due to issues in the current proof code, this commit also disables proof
checking for five QF_LRA benchmarks (see issue #2855).
|
|
This commit adds a check to make sure that the result of a `(check-sat)`
call matches the expected result set via `(set-info :status ...)`. In
doing so, it also fixes an issue where CVC4 would crash if asked for the
unsat core after setting the status to `unsat` but before calling
`(check-sat)` (see regression for concrete example). This happened
because CVC4 was storing the expected result and the computed result
both in the same variable (the expected result wasn't really being used
though). This commit keeps track of the expected result and the computed
result in separate variables to fix that issue.
|
|
When printing an empty symbol name, which can appear in an SMT2 file as
`||`, we were printing the empty string instead of quoting the symbol.
This commit fixes the issue and adds a regression test.
|
|
Fixes #3020. Boolean terms that appear in other terms, e.g. a Boolean
array index, are replaced by `BOOLEAN_TERM_VARIABLE`s to make sure that
they are handled properly in theory combination. When doing this
replacement, an equality of the form `(= <Boolean term> <Boolean term
variable)` is added to the assertions. The problem was that
`Theory::ppAssert()` would derive a substitution when this equality was
registered. The commit fixes the problem by not allowing to add
substitutions for `BOOLEAN_TERM_VARIABLE`s.
|
|
Fixes #2932. fp.roundToIntegral was rounding some very small subnormals up to
between 1 and 2, which is A. wrong and B. not idempotent. The
corresponding symfpu update fixes this as it was an overflow caused
by the unpacked significand not being able to represent an extra
significand bits.
|
|
|