The DOI matches its OpenAlex record.
Tolerating inconsistency
Abstract
This is a short paper about a simple technique for making applications robust by marking data that violates specially identified invariants and using those marks as guards around code which is sensitive to that invariant.Such a capability is needed when dealing with the exceptions that arise in practical applications and distinguish them from idealized specifications. Previous techniques either required the removal of the constraint so that the exception would be allowed, or its modification to exclude the particular exception from the restriction.Our technique softens the entire constraint without introducing special cases, and is intended primarily for temporary exceptions. Rather than synchronizing all updates required by some constraint, it allows them instead to occur distributed in time in any arbitrary order. Thus, it is particularly useful in multiclient applications with shared responsibilities and authority, such as cooperative design (including software development).While the constraint is violated, the offending data is automatically marked by guards to screen it from, or identify it to, code segments sensitive to the violation. When the time distributed update is complete so that the data is no longer inconsistent, the guards are automatically removed so that the data can again be processed normally.The relevance of this work to the Process Workshop is that it provides a technique for relaxing the synchronization problem among multiple users cooperating on some compound update and allows them instead to operate asynchronously. That is, this technique allows us to write more realistic process programs describing the flexible time-distributed coordinations among agents without loosing the formal constraints on their shared goals. The cost of this flexibility is the conditionalization of the cooperating agents to tolerate these temporary inconsistencies.We have created two implementations of our technique for softening constraints to allow temporary inconsistencies. Both are fully automatic and both handle arbitrary constraints expressed in our first-order logic formalism. The first converts the original constraint into a pair of constraints that require the presence of the guards while the original constraint is violated and their absence otherwise. These new constraints have “repairs” which fix violations by respectively inserting or removing the guards. The second implementation replaces the original constraint by an inference rule which derives the guard if and only if the original constraint predicate was violated.Our contribution lies in formally representing constraints whose temporary inconsistency can be tolerated by an application, and in automatically marking the particular offending data while the constraint is being violated. The goal of minimizing the number and duration of such violations is informal and implicit. The task of identifying sensitive code segments and determining how they should respond to the temporary inconsistencies is likewise informal and manual.Our work has an interesting relationship with the recent work on transactions by Pu, Kaiser, and Hutchinson @cite(Split-Transactions) addressing the times and set of users over which a constraint applies. They, too, are interested in multi-party transactions, and have proposed “splitting” transactions into pieces with differing visibility scopes so that users in some group could see the changes made by the splits within the transaction, while those outside the group wouldn't see any changes until the entire transaction was committed.Split-Transactions can be viewed as another approach to exception handling. Rather than dealing with an exception or an inconsistency after a completed transaction, Pu, Kaiser, and Hutchinson subdivide the transaction so that the multi-party coordinations can occur within it. As in all transactions, constraints are only applied when committed. Thus, during the transaction the effect is to remove the constraints.The power of the Split-Transaction approach is the scoping (potentially multi-level) that it provides to allow groups to cooperatively prepare a compound update for a broader community. However, within the transaction, since the constraints are not in effect, the inconsistency is not explicit, and hence cannot be tracked, managed, or supported except by informal mechanisms outside the system.But there is no reason why Split Transactions couldn't be combined with our own techniques for tolerating inconsistency to provide the advantages of both. While the inconsistency existed, it would be guarded by Pollution Markers, but it would only be visible to the group (dynamically formed) needed to resolve that inconsistency. After it was resolved, the Pollution Markers would be removed, and the new consistent state would be visible outside the group.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cited by
A framework for expressing the relationships between multiple views in requirements specification
Nuseibeh, B., Kramer, J., & Finkelstein, A. (1994) · IEEE Transactions on Software Engineering · doi:10.1109/32.328995 · Cited by 540
The three dimensions of requirements engineering: A framework and its applications
Pohl, K. (1994) · Information Systems · doi:10.1016/0306-4379(94)90044-2 · Cited by 269
Answering conceptual queries with Ferret
de Alwis, B., & Murphy, G. C. (2008) · Proceedings of the 30th international conference on Software engineering · doi:10.1145/1368088.1368092 · Cited by 63
Metadata
- Authors
- Balzer, R. M.
- Published
- 1991-05-01
- Pages
- 158-165
- DOI
- 10.5555/256664.256748
- OpenAlex
- W2294967233
- Cited by
- 237
- References
- 0
- FWCI
- 6.05
- Type
- conference paper
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Tolerating inconsistency. https://app.cochat.ai/papers/10.5555/256664.256748
- The paper
- Balzer, R. M. (1991). Tolerating inconsistency. https://doi.org/10.5555/256664.256748
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.