Age | Commit message (Collapse) | Author |
|
** file-level documentation at the top of the sources. **
This is the "make bugzilla stop bugging me" bugfix commit.
* Remove BackedNodeBuilder<> and collapse NodeBuilder<> hierarchy.
Updated documentation in the file. Resolves bug #99.
* Convenience NodeBuilders (PlusNodeBuilder, OrNodeBuilder, etc.)
moved into a separate file. Partially resolves bug #100.
* Moved isAssociative(Kind) into kind.h (and into the CVC4::kind
namespace) instead of metakind.h (where it was in CVC4::metakind).
This clears up a warning (private #inclusion) from the SMT and SMT2
parsers, and maybe makes more sense anyways, since this is based on
the kind (and not the metakind) of an operator.
* Documentation improvement; doxygen top-level \file gestures, \brief
gestures for files, etc. Changed contrib/update-copyright.pl for
this change, and post-processed to add \brief. Resolves bug #98.
* Removed ExprManager::mkExpr(Kind) and NodeManager::mkNode(Kind).
They no longer made sense. Resolves bug #91.
|
|
assignment for slack variables once solving has begun. (They cannot just be 0.) The second has to do with how assignments are backttacked. Assignments are now tracked all of the time, and are frozen once they are known to be consistent, i.e. after a successful updateInconsistentVars(). Also added a fuzz test that shows both of these problems to the regressions.
|
|
|
|
|
|
debugging statements. There is some evidence that these debugging statements were 20% of the running time for QF_LRA/miplib/fixnet-1000.smt in debug mode.
|
|
branch is a boolean should now be working. This fixes bug 138.
|
|
before being used. Fixed a bug in node_builder.h's crop where a pointer is dereferenced after a realloc call.
|
|
the tableau now simulates older pivots while adding a new row.
|
|
QF_LRA is now no longer complaining about seeing nodes that can be rewritten to CONST_BOOLEAN.
|
|
false. This is temporary and will be removed once TheoryEngine rewriting is more fully debugged.
|
|
|
|
debugging utility that prints out the lower bound, upper bound, assignment, and the constraints that were asserted that caused the lower bound and upperbound to be asserted.
|
|
for it as well. make check should now work again.
|
|
to its final destination. Added a basic rewriter to TheoryUF. (x=x rewrites to true.) Added DIVISION to src/expr/node_manager.cpp's getType. Fixed the theory returned for variables in theoryOf() for bool and arith. Fixed TheoryEngine rewrite children to properly handle APPLY_UFs. Removed the special case for equality in TheoryEngine rewrite. A few tests are currently broken due to deallocation errors. Morgan and I will try to commit a fix to this in a little bit.
|
|
potential design for later. Resolves bug 113, invalidates bugs 93 and 94.
|
|
support has been added. Also a few bug fixes to Tableau.
|
|
|
|
rewriter.
|
|
Rational and Integer constructors are now explicit. This means that 'Integer = 3;' and so on are no longer permitted. This closes bug 121.
|
|
To be done:
* Add erasable map Clause* to bool to minisat (backtracks with the solver)
* Add map from Clause* to Node (clauses that came from a node)
* Add reference counting Literal -> Node to CNFManager
* If Literal -> Node refcount goes to zero, the clauses of Node can be erased (if eresable)
* If clause is erased for each L in clause L -> Node refcount goes down
|
|
Real. See Bug 127 for a discussion of the hack. I am also adding a regression test that does not work (bug 128). It is not enabled so make check should still be fine.
|
|
TheoryArith. Also directly hacked in support for theoryOf() to work for equalities where the left hand is a variable of type real.
|
|
fixes and organizational changes. The theory is now enabled to perform checking in the TheoryEngine. This draft can now solve 2 new regression tests test/regress/regress0/ineq_slack.smt and test/regress/regress0/ineq_basic.smt. There is also a small bug fix inside src/expr/attribute.h.
|
|
|
|
|
|
|
|
|
|
instead of the kind, i.e.
Expr op = ..."f"...
em.mkExpr(op, children);
Operator kinds are automatically associated with the enclosing expression kind in the DSL and generated.
|
|
This should not affect other parts of the system. This code is untested, and is volatile. It is going to be improved in the future so don't pick on it too much. I am merging this code in its current state back into the main trunk because it was getting unruly to keep merging the updated trunk back into the branch.
|
|
Also a bug fix to a couple of unit tests.
|
|
* Types are now represented as nodes in the attribute table and are managed, i.e. you can say
Type booleanType = d_nodeManager->booleanType();
Type t = d_nodeManager->mkFunctionType(booleanType, booleanType);
FunctionType ft = (FunctionType)t;
Assert(ft.getArgTypes()[0], booleanType);
* The attributes now have a table for Nodes and a table for TNodes (both should be used with caution)
* Changes the way nodes are extracted from NodeBuilder, added several methods to
extract a Node, NodeValue, or Node*, with corresponding methods for extraction
* Used the above in the construction of Expr and Type objects
* The NodeManager now destroys the attributes in the destructor by pausing the
garbage collection
* To achive destruction a flag d_inDesctruction has been added to loosen the assertion
in NodeValue::dec() (there might be -refcount TNodes leftover)
* Beginnings of the Bitvector constants using GMP
Not yet in tiptop phase, needs more documentation, and Types should be pulled out to TypeNodes eventually. Also, the types are currently defined in the builting_kinds, and I need to add these to the theory specific definitions with special 'type' constructs.
I hate branching and merging.
|
|
auto-generated headers (metakind.h etc.), so they don't have to be
recompiled every time. This drastically improves build time when
only small updates are made.
* Added "memory.h" unit test header for checking out-of-memory
conditions. cdlist_black uses it.
* Added helpful output when you "make lcov" in a non-coverage-enabled
build.
* Removed strict aliasing warning when compiling metakind.h header
with optimization on.
* Removed const version of NodeBuilder::operator Node()---it was
poorly performing, better to not permit it---and fixed the
convenience builders to use the non-const version
(re: code review #63)
* Color-coded test output on capable terminals.
* Fixed some warnings in unit tests.
|
|
|
|
+ Context* getContext() -- gets the context
+ ContextMemoryManager* getCMM() -- gets the CMM
+ int getLevel() -- the scope level of the ContextObj's most recent update
+ bool isCurrent() -- true iff the most recent update is the current top level
In particular, the ContextObj::getCMM() call cleans up by TheoryUF's
ECData::addPredecessor() function substantially (re: code review bug #64).
* Fix serious bugs in context operations that corrupted the ContextObj
linked lists. Closes bug #85.
* Identified a bug in the way objects of the "Link" class are
allocated; see bug #96.
* Re-enable context white-box tests that ensure proper links in linked
lists. Closes bug #86.
* Re-enable CDMap<>::emptyTrash(). Closes bug #87.
* Add a tracing option (-t foo or --trace foo) to the driver to enable
Trace("foo") output stream. -d foo implies -t foo.
* Minor clean-up of some TheoryUF code; addition of some documentation
(re: code review bug #64).
* Address some things that caused Doxygen discomfort.
* Address an issue raised in NodeManager's code review (bug #65).
* Remove an inaccurate comment in Attribute code (re: code review bug #61).
|
|
instead of assuming it's atomic based on kind. Atomicity is
determined at node building time. Fixes bug #81. If this is
determined to make node building too slow, we can allocate another
attribute "AtomicHasBeenComputed" to lazily compute atomicity.
* TheoryImpl<> has gone away. Theory implementations now derive from
Theory directly and share a single RegisteredAttr attribute for term
registration (which shouldn't overlap: every term is "owned" by
exactly one Theory). Fixes bug #79.
* Additional atomicity tests in ExprBlack unit test.
* More appropriate whitebox testing for attribute ID assignment
(AttributeWhite unit test).
* Better (and more correct) assertion checking in NodeBuilderBlack.
* run-regression script now checks exit status against what's provided
in "% EXIT: " gesture in .cvc input files, and stderr against
"% EXPECT-ERROR: ". These can be used to support intended failures.
Fixes bug #84. Also add "% EXIT: " gestures to all .cvc regressions
in repository.
* Solved some "control reaches end of non-void function" warnings in
src/parser/bounded_token_buffer.cpp by replacing
"AlwaysAssert(false)" with "Unreachable()" (which is known
statically to never return normally).
* Regression tests now use the cvc4 binary under
builds/$(CURRENT_BUILD)/src/main instead of the one in bin/ which
may not be properly installed yet at that point of the build.
(Partially fixes bug #46.)
* -fvisibility=hidden is now included by configure.ac instead of each
Makefile.am, which will make it easier to support platforms
(e.g. cygwin) that do things a different way.
* TheoryUF code formatting. (re: my code review bug #64)
* CDMap<> is leaking memory again, pending a fix for bug #85 in the
context subsystem. (To avoid serious errors, can't free context
objects.)
* add ContextWhite unit test for bug #85 (though it's currently
"defanged," awaiting the bugfix)
* Minor documentation, other cleanup.
|
|
Fix the case in NodeBuilderBlack that triggered bug #82. (Fixes bug #82.)
This also fixes regression failures from this morning (2010 Apr 4), in
the optimized builds, for which a fix was included in 365 and reverted
in 375. They looked like this:
In ExprBlack::testGetConst:
/usr/local/share/cvc4/src/cvc4-2010-04-04/builds/x86_64-unknown-linux-gnu/production/../../../test/unit/expr/expr_black.h:377: Error: Expected (a->getConst<Kind>()) to throw (IllegalArgumentException) but it didn't throw
/usr/local/share/cvc4/src/cvc4-2010-04-04/builds/x86_64-unknown-linux-gnu/production/../../../test/unit/expr/expr_black.h:378: Error: Expected (b->getConst<Kind>()) to throw (IllegalArgumentException) but it didn't throw
[etc..]
|
|
|
|
comments to files without them
|
|
Type code review and closed bug #25.
* Do assertions on Expr creation (public library interface) even when
assertions are off. Also a similar check for proper kind (in public
interface) when Expr::getConst<>() is called. This fixes a unit
test that was failing in production builds (an exception wasn't
thrown but should have been).
* kind::XOR must have exactly 2 arguments, not 2-or-more.
|
|
* Other minor changes to the new parser to match coding guidelines,
add documentation, ....
* Add CFLAGS stuff to configure.ac parser Makefile.ams. This ensures
that profiling, coverage, optimization, debugging, and warning
level options will apply to the new parser as well (which is in C,
not C++). This fixes the deprecated warning we were seeing this
evening.
* Now, if you have ANTLR_HOME set in your environment, you don't need
to specify --with-antlr-dir to ./configure or have libantlr3c
installed in standard places. --with-antlr-dir still overrides
$ANTLR_HOME, and if the installation in $ANTLR_HOME is missing or
doesn't work, the standard places are still tried.
* Extend "silent make" to new parser stuff.
* Added src/parser/bounded_token_buffer.{h,cpp} to the list of
exclusions in contrib/update-copyright.pl and mention them as
excluded from CVC4 copyright in COPYING. They are antlr3-derived
works, covered under a BSD license.
OTHER STUFF:
* expr_manager.h, expr.h, expr_manager.cpp, and expr.cpp are now
auto-generated by a "mkexpr" script. This provides the correct
instantiations of mkConst() for public use, e.g., by the parser.
* Fix doxygen documentation in expr, expr_manager.. closes bug #35
* Node::isAtomic() implemented in a better way, based on theory kinds
files. Fixes bug #40. To support this, a "nonatomic_operator"
command has been added. All other "parameterized" or "operator"
kinds are atomic.
* Added expr_black test
* Remove kind::TRUE and kind::FALSE and make a new CONST_BOOLEAN kind
that takes a "bool" payload; for example, to make "true" you now do
nodeManager->mkConst(true).
* Make new "cvc4_public.h" and "cvc4parser_public.h" headers. Private
headers should include "cvc4_private.h"
(resp. "cvc4parser_private.h"), which existed previously. Public
headers should include the others. **No one** should include the
autoheader #include (which has been renamed "cvc4autoconfig.h")
directly, and public CVC4 headers can't access its #defines. This
is to avoid us having the same distribution problem as libantlr3c.
* Preliminary fixes based on Tim's code review of attributes (bug #61).
This includes splitting hairy template internals into
attribute_internals.h, for which another code review ticket will be
opened. Bug is still outstanding, but pending further
refactoring/documentation.
* Some *HashFcns renamed to *HashStrategy to match refactoring done
elsewhere (done by Chris?) earlier this week.
* Simplified creation of make rules for generated files (expr.cpp,
expr.h, expr_manager.cpp, expr_manager.h, theoryof_table.h, kind.h,
metakind.h).
* CVC4::Configuration interface and implementation split (so private
stuff doesn't leak into public headers).
* Some documentation/code formatting fixes.
* Add required versions of autotools to autogen.sh.
* src/expr/mkmetakind: fix a nonportable thing in invocation of "expr"
that was causing warnings on Red Hat.
* src/context/cdmap.h: add workaround to what appears to be a g++ 4.1
parsing bug.
|
|
|
|
|
|
|
|
* add NodeManagerWhite unit test
* change kind::APPLY to kind::APPLY_UF
* better APPLY handling: operators are no longer considered children
* more efficient pool lookup; the NodeValue doesn't have to be as fully constructed for the lookup to proceed
* extend DSL for kind declarations
+ new "theory" command declares a theory and its header. theory_def.h no longer required.
+ arity enforced on operators
+ constant mapping, hashing, equality
* CONSTANT metakinds supported (via Node::getConst<T>(), for example, Node::getConst<CVC4::Rational>() gets a Rational out of a Node (assuming it is of CONST_RATIONAL kind)
* added CONST_RATIONAL and CONST_INTEGER kinds
* builtin operators (AND, OR, PLUS, etc..) returned by Node::getOperator() are now CONSTANT metakind and are created by NodeManager
* Pretty-printing of Nodes now has a depth limit settable by a stream manipulator (e.g. "cout << Node::setdepth(5) << m << endl;" prints DAG rooted at m to a depth of 5)
* getters added to Node, TNode, NodeValue, etc., for operators and metakinds
* build-time generators for kind.h, metakind.h, and theoryof_table.h headers now have a simpler design and flag errors better, and the templates (kind_template.h etc.) are easier to understand.
* DISTINCT is now a kind, and the SMT parser now passes through DISTINCT nodes instead of blowing them up into ANDs. Until theory rewriting is online, though, DISTINCTs are directly blown up into conjunctions in TheoryEngine::rewrite().
* add gmpxx detection and inclusion
* better Asserts throughout, some documentation, cleanup
|
|
Minor name changes for cleanup and hash function templates
|
|
different "kinds of kinds" (special, operator, parameterized, and constant), and permits doxygen comments on them
|
|
CDList, and CDMap. Added the d_underTheShotgun field to NodeManager to keep track of which NodeValue is currently being deleted. If a Node or TNode has this node value, it can always be deleted. This avoids the need for introducing SoftNodes. Currently passes Debug and Production make check
|
|
test/unit/expr/node_white.h: add whitebox attribute test (pulled out
attribute stuff from node_white)
* test/unit/parser/parser_black.h: fix memory leaks uncovered by
valgrind
* src/theory/interrupted.h: actually make this "lightweight" (not
derived from CVC4::Exception), as promised in my last commit
* src/theory/uf/theory_uf.h, test/unit/expr/attribute_black.h: match
the new-style cleanup function definition
* src/expr/attribute.cpp, src/expr/attribute.h: support for attribute
deletion, custom cleanup functions, clearer cleanup function
definition.
* src/expr/node_manager.h, src/expr/node_manager.cpp: reclaim
remaining zombies in dtor, rename NodeValueSet ==> "NodeValuePool",
and enable freeing of NodeValues
* src/expr/type.h, src/expr/type.cpp: reference-counting for types,
customized cleanup function for types, also code cleanup
* (various): changed "const Type*" to "Type*" (to enable
reference-counting etc. Types are still immutable.)
* src/util/output.h: add ::isOn()-- which queries whether a
Debug/Trace flag is currently on or not.
* src/smt/smt_engine.cpp, src/parser/antlr_parser.cpp,
src/expr/type.cpp, src/expr/expr_manager.cpp, various others:
minor code cleanup
|
|
* CVC4::theory::Interrupted no longer derives CVC4::Exception.
* Interrupted is only thrown if "safe" parameter is TRUE !
* UF returns one conflict (instead of waiting for Interrupted to be thrown).
* Minor build system work (quieter builds if V=0, better handling of build
profiles in configure)
|
|
with old iterator (closes bug #47).
* src/context/cdset.h: implemented.
* src/expr/node_builder.h: fixed all the strict-aliasing warnings.
* Remove Node::hash() and Expr::hash() (they had been aliases for
getId()). There's now a NodeValue::internalHash(), for internal
expr package purposes only, that doesn't depend on the ID. That's
the only hashing of Nodes or Exprs.
* Automake-quiet generation of kind.h, theoryof_table.h, and CVC and
SMT parsers.
* various minor code cleanups.
|