Repository navigation
Unit tests of NotAllEqual - #130
Merged
Merged
Conversation
The function NotAllEqual() of the section Constraints (Global) is tested as the previous global constraints: forms of the arguments, sizes, domains, holes, repeated variables, degenerated cases, expressions (with and without -mini), integers among the terms, groups, logical contexts, XCSP3 files and invalid cases, each constraint being solved with ACE, CHOCO and cosoco and compared with brute force. The known bugs are marked as expected failures, to be reported.
This was referenced Sep 28, 2026
Closed
…Equal The provisional marks '(to be reported)' now give the issues: #131, #132 and #133 for PyCSP3, xcsp3team/ACE#18 and xcsp3team/ACE#19 for ACE, and section 5 of chocoteam/choco-solver#1248 for CHOCO. The mark of the empty tables of supports (ACE and CHOCO) is kept as in the other test files.
szczepanskiNicolas
added
run-tests
NValues() returned a constant for a single term before checking its type, so
that NotAllEqual() given None, a string, a float, a range or a dictionary
posted the constraint false (with a warning), and NValues('a') == 1 was
accepted without any message. None is now reported by NValues() and
NotAllEqual(), and the shortcut for a single term only applies to a variable,
an expression, a partial constraint or an integer: an invalid term is reported
by checkType(), as for AllEqual().
Fixes #131
…erent() and AllEqual() NotAllEqual() given only integers replaced each one by an auxiliary variable (NotAllEqual(1, 2) was always true, NotAllEqual(1, 1) always false), or posted the constraint false (NotAllEqual(5)). As for AllDifferent() and AllEqual(), an explicit error is now reported; integers among variables remain accepted. The terms are flattened once: the body of NValues() is moved to _nvalues(), which receives a flat list, and is called by NValues() and NotAllEqual(), so that the check does not cost a second call to flatten() (+16% on the generation of 23,490 constraints NotAllEqual, against +1.5% now). Fixes #132
nValues is not part of the perimeter of the mini-tracks of the XCSP competitions (intension, extension, allDifferent, sum and element), so that the option -mini has nothing to change for NValues() (#133, not a bug): the test expecting auxiliary variables in <nValues> with -mini was wrong, and the test solving the constraints with -mini repeated the one without -mini, the generated files being the same.
A term given several times does not change the number of distinct values; ACE fails on a variable given several times in nValues and cosoco refuses it, so that PyCSP3 is expected to keep each term once before posting the constraint (#134; xcsp3team/ACE#19 closed for this reason). The constraints with a repeated variable, or with two identical sums replaced by the same auxiliary variable, are solved with at least three distinct terms (with two terms, ACE fails because of xcsp3team/ACE#18), and the generated elements <nValues> are checked; NotAllEqual(x[0], x[0]) is expected to be trivially false.
…he constraint A term given several times does not change the number of distinct values, but ACE fails on a variable given several times in nValues and cosoco refuses it. Before posting the constraint, _nvalues() now keeps each term once: the variables are compared by identity (== being redefined for building expressions), the expressions by their textual forms, identical partial constraints being replaced by the same auxiliary variable, and an integer given several times is replaced by a single auxiliary variable. When a single term remains (NotAllEqual(x[0], x[0])), the constraint is trivially false, as for NotAllEqual(x[0]). On the generation of 23,490 constraints NotAllEqual, the time is unchanged on variables (1.43 s) and 3.5% longer on expressions (3.60 s -> 3.73 s), with identical files. Fixes #134
This was referenced Sep 29, 2026
This was referenced Oct 15, 2026
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
2 participants
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.
Unit tests of the constraint NotAllEqual of the section Constraints (Global) of the API (https://pycsp.org/documentation/api/constraints-global/#NotAllEqual), written as for the previous global constraints (#66), with the same harness.
This PR adds tests (
pycsp3/tests/test_constraints_global_notallequal.py). The bugs found are reported in issues, and the tests concerned are marked as strict expected failures with the numbers of these issues. It also contains the fixes of #131 (9327cec) and #132 (57e5479), which remove the marks of these issues, the removal of the tests with the option-mini(baeef7e), #133 not being a bug, and the tests of the terms given several times, expected to be kept once (41f5cdd), with the fix of #134 (55439f8).Tests
NotAllEqual()postsNValues(...) > 1, that is to say an element<nValues>with the condition(gt,1). The tests cover:NotAllEqual()in logical expressions (~,|,&,imply(),either());Each constraint is solved with ACE, CHOCO and cosoco, and all the solutions are compared with the ones computed by brute force.
test_constraints_global_notallequal.pytest_constraints_global_notallequal.pyWhole suite at 55439f8 (run locally, master not having moved since the creation of the branch): 2703 passed, 125 skipped and 203 expected failures, no failure (master at 0f4ffd2 plus this file).
Known bugs
Each issue has the labels
bug,unit-tests,Constraints (Global)and the label of the function (NotAllEqualfor PyCSP3,nValuesfor ACE). Each issue of a solver shows the results of the three solvers on the same XCSP3 file, compared with brute force.Expected failures at the last commit (20 tests):
nValueswith(gt,1)or(ge,2)on two variables fails; fixed in ACE (3a6cd59) and closed, the marks being kept as long as PyCSP3 embeds ACE 2.6 (with ACE built from 3a6cd59, these 6 tests pass)nValueson expressions, some solutions are found several timesbuildCtrFalse()(Constraint with only conflicts, not reported, by decision, see #128)Issues of PyCSP3 handled in this PR:
NotAllEqual()with a single invalid term (None, a string, a range...) posts false instead of reporting an errorNotAllEqual()given only integers is accepted, whileAllDifferent()andAllEqual()report an error-mini,NValues()keeps the expressions in its list instead of replacing them by auxiliary variablesnValuesis not part of the mini-tracks (intension, extension, allDifferent, sum and element); closed as not planned, and the tests with-miniremoved in baeef7eNotAllEqual()keeps the terms given several times, on which ACE fails (and that cosoco refuses); first reported as xcsp3team/ACE#19, closed because each term is to be kept once by PyCSP3Cases not handled by the solvers
The following cases are not reported as bugs, because the solver says that it does not handle them (or, for the symbolic variables, because they are not part of XCSP3-core); the corresponding tests are skipped with this solver (11 tests). The other cells give the result of the solver: handled (✓), or a bug reported in the given issue.
nValueson expressionss UNSUPPORTED)nValueson symbolic variables (not part of XCSP3-core)RuntimeExceptionRuntimeException: UNSUPPORTED