# test impact analysis

Test impact analysis is a software-testing technique that maps code changes to affected tests and runs only the relevant subset.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## CircleCI Expands Smarter Testing to Power Independent Validation for AI-Driven Development

DevFeed: [CircleCI Expands Smarter Testing to Power Independent Validation for AI-Driven Development](<https://devfeed.tech/articles/circleci-expands-smarter-testing-to-power-independent-validation-for-ai-driven-development-144327.md>)

Original publisher: [Read original article](<https://sdtimes.com/software-testing/circleci-expands-smarter-testing-to-power-independent-validation-for-ai-driven-development/>)

Author: SD Times Newswire

Published: 2026-11-19T18:02:50Z

Content type: news

Language: en

Sources: [SD Times](<https://devfeed.tech/sources/sd-times.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [CI/CD Pipeline](<https://devfeed.tech/topics/ci-cd-pipeline.md>), [CircleCI](<https://devfeed.tech/topics/circleci.md>), [auto rerun failed tests](<https://devfeed.tech/topics/auto-rerun-failed-tests.md>), [Continuous integration](<https://devfeed.tech/topics/continuous-integration.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [ai-powered](<https://devfeed.tech/tags/ai-powered.md>), [ai-powered-customer-service](<https://devfeed.tech/tags/ai-powered-customer-service.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [circleci](<https://devfeed.tech/tags/circleci.md>), [continuous-delivery](<https://devfeed.tech/tags/continuous-delivery.md>), [continuous-integration](<https://devfeed.tech/tags/continuous-integration.md>), [development](<https://devfeed.tech/tags/development.md>), [dymanic-parallelism](<https://devfeed.tech/tags/dymanic-parallelism.md>), [dynamic-test-splitting](<https://devfeed.tech/tags/dynamic-test-splitting.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [faster](<https://devfeed.tech/tags/faster.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [latest-news](<https://devfeed.tech/tags/latest-news.md>), [newswire](<https://devfeed.tech/tags/newswire.md>), [parallelism](<https://devfeed.tech/tags/parallelism.md>), [platform](<https://devfeed.tech/tags/platform.md>), [smarter-testing](<https://devfeed.tech/tags/smarter-testing.md>), [testing](<https://devfeed.tech/tags/testing.md>), [tests](<https://devfeed.tech/tags/tests.md>), [validation](<https://devfeed.tech/tags/validation.md>)

### AI overview

CircleCI's Smarter Testing suite aims to speed validation and reduce CI compute by running tests affected by a code change and adjusting machine count to match selected work. Developers and AI coding agents can run affected tests locally before pushing at no additional charge; in CI, Intelligent Test Selection selects tests, Dynamic Parallelism adjusts machines, Dynamic Test Splitting balances work, and Auto Rerun retries only failed tests. CircleCI says the suite is built in, with no additional vendor or infrastructure required. Customers report different outcomes: Medallion cites lower cost per pull request and shorter builds as pull request volume grew 10x and its suite 2.5x; Kustomer says test costs fell by a quarter and feedback became faster. These are customer-reported results, not a guarantee of similar savings.

### Source excerpt

SAN FRANCISCO, CA -- CircleCI, a leading continuous integration and continuous delivery (CI/CD) platform, announced new features available in its Smarter Testing suite. Smarter Testing helps engineering teams get faster feedback by running only the tests affected by a code change, and new Dynamic Parallelism adjusts the number of machines for each run to cut... ... continue reading The post CircleCI Expands Smarter Testing to Power Independent Validation for AI-Driven Development appeared first on SD Times.

## AI Code Review Requires Revision-Aware Evidence of Concern Resolution

DevFeed: [AI Code Review Requires Revision-Aware Evidence of Concern Resolution](<https://devfeed.tech/articles/ai-code-review-requires-revision-aware-evidence-of-concern-resolution-142147.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23275>)

Author: Mamdouh Alenezi

Published: 2026-11-19T05:00:00Z

Content type: opinion

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [ai code review](<https://devfeed.tech/topics/ai-code-review.md>), [Code review](<https://devfeed.tech/topics/code-review.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [code-review](<https://devfeed.tech/tags/code-review.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [revision](<https://devfeed.tech/tags/revision.md>), [test](<https://devfeed.tech/tags/test.md>), [verification](<https://devfeed.tech/tags/verification.md>)

### AI overview

Closing a review thread or changing a commented line is not evidence that a concern is fixed; later revisions may undo a repair. The paper proposes treating concerns as persistent verification obligations, with REVAL rechecking evidence when its dependencies change. In synthetic histories, REVAL matched exhaustive reassessment across 6,420 paired comparisons while reducing post-comment test executions by 34%. Its false-resolution rate and recurrence detection varied with how witnesses were selected: 0.11 and 80.4% with three oracle-sampled witnesses, versus 0.13 and 88.3% using witnesses from review comments and independent held-out checks. A sticky outdated-line proxy falsely called 72.9% of unresolved states resolved, and stopping after initial success missed all recurrences. The controlled results distinguish revalidation correctness from witness adequacy; maintenance overhead made REVAL slower for microsecond-scale tests, with an estimated break-even test cost of 0.42 ms.

### Source excerpt

arXiv:2611.23275v1 Announce Type: new Abstract: AI-assisted code review increasingly judges whether reported concerns have been resolved. Yet closing a thread or modifying a commented line does not establish repair, and later revisions can invalidate earlier fixes. This position paper argues that resolution requires inspectable, revision-specific evidence that is revalidated when its dependencies change. We model review concerns as persistent verification obligations, derive a sufficient condition under which dependency-gated revalidation matches exhaustive reassessment, and introduce REVAL, a lightweight procedure adapting principles from regression-test selection and build systems. We evaluate REVAL against oracle ground truth in synthetic revision histories covering 441 concerns, 1,284 assessments, and 12 defect families. A sticky outdated-line proxy falsely marked 72.9% of unresolved states as resolved, while stopping checks after initial success missed every recurrence. With three oracle-sampled witnesses, REVAL achieved a false-resolution rate of 0.11 (95% CI: 0.06-0.16) and detected 80.4% of recurrences. With witnesses derived solely from review comments and assessed using independent held-out checks, the corresponding results were 0.13 and 88.3%. Across 6,420 paired comparisons, REVAL matched exhaustive reassessment while reducing post-comment test executions by 34%. Maintenance overhead increased runtime for microsecond-scale tests; the estimated break-even test cost was 0.42 ms. These findings support revision-aware evidence maintenance in a controlled setting, while distinguishing revalidation correctness from witness adequacy. A prespecified evaluation protocol outlines how to assess generalisability in real review histories.

## Predictive Test Optimization Without Historical Failure Data

DevFeed: [Predictive Test Optimization Without Historical Failure Data](<https://devfeed.tech/articles/predictive-test-optimization-without-historical-failure-data-137960.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18574>)

Author: Maximilian Jungwirth, Raphael N\\"ommer, Andreas Stahlbauer, Sven Apel, Gordon Fraser

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>)

Tags: [cs-se](<https://devfeed.tech/tags/cs-se.md>), [flaky-tests](<https://devfeed.tech/tags/flaky-tests.md>), [industry](<https://devfeed.tech/tags/industry.md>), [labeling](<https://devfeed.tech/tags/labeling.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [pipelines](<https://devfeed.tech/tags/pipelines.md>), [test-optimization](<https://devfeed.tech/tags/test-optimization.md>)

### AI overview

Co-evolution labeling predicts which tests are relevant to a code change by using tests and code that change together in version history, avoiding reliance on historical test results and reducing exposure to flaky-test label noise. Evaluated on three industry and open-source projects, each with thousands of tests and over 1 million lines of code, it nearly matched failure-history-trained models' failure detection, with an overall Average Percentage of Faults Detected of 0.88. The excerpt does not provide further evaluation details.

### Source excerpt

arXiv:2611.18574v1 Announce Type: new Abstract: As software systems grow in size and complexity, their test suites become increasingly expensive to execute, delaying feedback and wasting resources. State-of-the-art predictive test optimization methods use machine learning to select tests relevant to a change, typically trained on historical failures. This depends on failure histories and is susceptible to label noise from flaky tests. We introduce the novel method of co-evolution labeling for predictive test optimization, deriving test relevance from tests and code changing together in the version history. These data are widely available and less noisy. We train and evaluate models with failure and co-evolution labels on three large projects---from industry and open source; each with thousands of tests and more than 1M LOC. Co-evolution labeling nearly matches the failure detection capabilities of failure-based models, with an overall Average Percentage of Faults Detected of 0.88, while being more resistant to label noise and requiring no test result history. Co-evolution labeling offers a practical, immediately deployable path to faster pipelines and reliable predictive test optimization.

## Test Automation Architecture: Components, Design Techniques, and Implementation

DevFeed: [Test Automation Architecture: Components, Design Techniques, and Implementation](<https://devfeed.tech/articles/test-automation-architecture-components-design-techniques-and-implementation-136321.md>)

Original publisher: [Read original article](<https://testgrid.io/blog/test-automation-architecture/>)

Author: Tejas Patel

Published: 2026-11-16T13:26:54Z

Content type: article

Language: en

Sources: [TestGrid](<https://devfeed.tech/sources/testgrid-blog.md>)

Topics: [Test automation](<https://devfeed.tech/topics/test-automation.md>), [software-architecture](<https://devfeed.tech/topics/software-architecture.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Page Object](<https://devfeed.tech/topics/page-object.md>), [Selenium](<https://devfeed.tech/topics/selenium.md>), [Playwright](<https://devfeed.tech/topics/playwright.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [architecture-components](<https://devfeed.tech/tags/architecture-components.md>), [automation](<https://devfeed.tech/tags/automation.md>), [components](<https://devfeed.tech/tags/components.md>), [core](<https://devfeed.tech/tags/core.md>), [data](<https://devfeed.tech/tags/data.md>), [design](<https://devfeed.tech/tags/design.md>), [environments](<https://devfeed.tech/tags/environments.md>), [execution](<https://devfeed.tech/tags/execution.md>), [guide](<https://devfeed.tech/tags/guide.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [reporting](<https://devfeed.tech/tags/reporting.md>), [techniques](<https://devfeed.tech/tags/techniques.md>), [test](<https://devfeed.tech/tags/test.md>), [test-automation](<https://devfeed.tech/tags/test-automation.md>), [test-automation-architecture](<https://devfeed.tech/tags/test-automation-architecture.md>)

### AI overview

Test automation architecture organizes test creation, scripts, data, application interfaces, execution infrastructure, and reporting so a suite can grow without excessive maintenance. The text describes four layers: generation, definition, execution, and adaptation. It recommends assessing the application and requirements, selecting test types and a framework, using reusable patterns such as page object model, preparing isolated environments and data, and integrating runs and quality gates into CI/CD. Data-driven frameworks separate inputs from scripts; keyword-driven frameworks assemble actions from predefined terms; hybrid frameworks combine approaches and reusable components. It names Selenium WebDriver, Playwright, Appium, and Cypress for different web and mobile testing needs, and describes AI uses including test generation, locator recovery, and regression prioritization. The supplied excerpt also presents TestGrid's product capabilities, including framework support, parallel device and browser execution, reporting, CI/CD integrations, and deployment options.

### Source excerpt

A test suite can look perfectly manageable at 50 tests and become painful at 500. Shared data starts colliding, environments drift, execution slows down, reporting gets noisy, and one UI change can ripple through dozens of scripts. This guide to test automation architecture shows you how to prevent that sprawl. You'll see how the core [...]

## How to Speed Up CI in the Age of AI-Written Code

DevFeed: [How to Speed Up CI in the Age of AI-Written Code](<https://devfeed.tech/articles/how-to-speed-up-ci-in-the-age-of-ai-written-code-133352.md>)

Original publisher: [Read original article](<https://www.freecodecamp.org/news/speed-up-ci-in-the-age-of-ai-written-code/>)

Author: Manish Shivanandhan

Published: 2026-11-13T12:46:22Z

Content type: article

Language: en

Sources: [Free Code Camp](<https://devfeed.tech/sources/freecodecamp-programming-tutorials-python-javascript-git-more.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [avrea](<https://devfeed.tech/tags/avrea.md>), [bazel](<https://devfeed.tech/tags/bazel.md>), [build](<https://devfeed.tech/tags/build.md>), [cache](<https://devfeed.tech/tags/cache.md>), [caching](<https://devfeed.tech/tags/caching.md>), [cargo](<https://devfeed.tech/tags/cargo.md>), [ci](<https://devfeed.tech/tags/ci.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [concurrency](<https://devfeed.tech/tags/concurrency.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [github](<https://devfeed.tech/tags/github.md>)

### AI overview

As AI agents increase the number of pull requests, CI queues and repeated work can erase the time saved writing code. The article recommends measuring queue, setup, and execution time first, then applying targeted fixes: reuse dependency and Docker caches with stable keys (while accounting for remote restore costs), run heavy jobs on faster hardware, parallelize independent jobs and shard tests, and skip unaffected or stale work. It compares runner options including Avrea, Depot, RunsOn, and GitHub larger runners, with different hosting and billing models; vendor performance claims and benchmark results should be checked against a team's own workflow. Sharding adds setup cost, required checks can remain pending when workflows use path filters, and cancelling in-progress runs is intended for ordinary CI, not deployments. The suggested approach is to change one thing at a time and remeasure.

### Source excerpt

Coding agents now write code faster than most CI pipelines can check it. In April 2026, GitHub COO Kyle Daigle said that his platform was handling 275 million commits a week, on pace for 14 billion in

## Caprese: Change Impact Recommendation Using History and Runtime Analysis

DevFeed: [Caprese: Change Impact Recommendation Using History and Runtime Analysis](<https://devfeed.tech/articles/caprese-change-impact-recommendation-using-history-and-runtime-analysis-131698.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.14937>)

Author: Sadjad Tavakoli, Saba Alimadadi

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Node.js](<https://devfeed.tech/topics/node-js.md>)

Tags: [analysis](<https://devfeed.tech/tags/analysis.md>), [applications](<https://devfeed.tech/tags/applications.md>), [artifacts](<https://devfeed.tech/tags/artifacts.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [change](<https://devfeed.tech/tags/change.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [dependency](<https://devfeed.tech/tags/dependency.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [events](<https://devfeed.tech/tags/events.md>), [execution](<https://devfeed.tech/tags/execution.md>), [framework](<https://devfeed.tech/tags/framework.md>), [function](<https://devfeed.tech/tags/function.md>), [functions](<https://devfeed.tech/tags/functions.md>), [history](<https://devfeed.tech/tags/history.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [impact](<https://devfeed.tech/tags/impact.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [javascript](<https://devfeed.tech/tags/javascript.md>), [recommendation](<https://devfeed.tech/tags/recommendation.md>), [runtime](<https://devfeed.tech/tags/runtime.md>)

### AI overview

Caprese ranks functions that may be affected by a change in Node.js code, combining historical co-change patterns with dependencies observed during execution. Its repository mode mines recurring, change-set-relevant patterns; its dynamic mode captures calls, callbacks, events, timers, and shared state; a hybrid mode integrates both. Across ten open-source applications and 100 selected change sets, 77.4% of relevant functions in the hybrid mode's top 60 were identified exclusively by one of the two analyses. Dynamic mode had the highest mean precision in the top five (0.65), while repository mode performed best at 60 recommendations (0.42). Hybrid mode beat TARMAQ on average but did not consistently surpass either individual mode.

### Source excerpt

arXiv:2611.14937v1 Announce Type: new Abstract: Change impact recommendation identifies program entities potentially affected by source code changes. Historical co-changes and runtime dependencies provide complementary evidence, particularly in JavaScript, where dynamic and asynchronous behaviour complicates dependency analysis. We present Caprese, an open-source framework for function-level change impact recommendation in Node.js. Given a change set, it produces a ranked impact set of potentially affected functions. It supports repository-based recommendation through confidence-guided mining of recurring, change-set-relevant co-change patterns of varying lengths; dynamic recommendation through execution-derived dependencies involving calls, callbacks, events, timers, and shared state; and hybrid recommendation integrating both analyses. Its modular implementation supports reusable analyses. Evaluation on ten open-source applications and 100 selected change sets shows that, on average across projects, 77.4% of the relevant functions among the top 60 hybrid recommendations are identified exclusively by one analysis, demonstrating substantial complementarity. Dynamic recommendation achieves the highest mean precision among the top five recommendations (0.65), while repository-based recommendation performs best when the ranking is extended to 60 recommendations (0.42). Hybrid recommendation outperforms TARMAQ on average but does not consistently outperform the individual modes. Caprese is publicly available with source code, documentation, and evaluation artifacts.

## Rethinking Requirement-to-Code Traceability Through Single-Agent and Multi-Agent Approaches

DevFeed: [Rethinking Requirement-to-Code Traceability Through Single-Agent and Multi-Agent Approaches](<https://devfeed.tech/articles/rethinking-requirement-to-code-traceability-through-single-agent-and-multi-agent-approaches-127211.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.09162>)

Author: Sarthak Banerjee, Sallam Abualhaija, Domenico Bianculli, Lionel Briand

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [development-process](<https://devfeed.tech/topics/development-process.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [change](<https://devfeed.tech/tags/change.md>), [classification](<https://devfeed.tech/tags/classification.md>), [code](<https://devfeed.tech/tags/code.md>), [cs-ce](<https://devfeed.tech/tags/cs-ce.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [maintenance](<https://devfeed.tech/tags/maintenance.md>), [multi-agent](<https://devfeed.tech/tags/multi-agent.md>), [precision](<https://devfeed.tech/tags/precision.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [recall](<https://devfeed.tech/tags/recall.md>), [regulatory-compliance](<https://devfeed.tech/tags/regulatory-compliance.md>), [repository](<https://devfeed.tech/tags/repository.md>), [requirements](<https://devfeed.tech/tags/requirements.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [source](<https://devfeed.tech/tags/source.md>), [traceability](<https://devfeed.tech/tags/traceability.md>), [variable](<https://devfeed.tech/tags/variable.md>), [verification](<https://devfeed.tech/tags/verification.md>), [windows](<https://devfeed.tech/tags/windows.md>)

### AI overview

The paper compares repository-wide LLM analysis with retrieval-based traceability for linking software requirements to code. A single agent analyzing the full repository reaches 80.9% recall, 38.6 percentage points above the retrieval baseline, but loses 25.8 points in precision. Multi-agent debate offers the best reported precision-recall balance: 24.5% precision, 71.0% recall, and an F2 score of 49.5%, outperforming the state of the art by 7.5 points and the single-agent approach by 2.7. The authors' qualitative analysis suggests retrieval works best when requirements echo code vocabulary, full-context single-agent reasoning when code is well documented and consistently named, and debate when evidence is indirect, conflicting, or distributed. Results vary considerably, so aggregate metrics alone provide limited explanation.

### Source excerpt

arXiv:2611.09162v1 Announce Type: cross Abstract: Requirements-to-code traceability supports essential software engineering activities, including change impact analysis, maintenance, and regulatory compliance. Existing approaches typically rely on a retrieval-and-classification pipeline, which suffers from an inherent retrieval bottleneck: relevant source files that are not retrieved cannot be recovered during subsequent analysis. In this paper, we revisit requirements-to-code traceability in light of the substantially larger context windows offered by modern large language models, which make repository-scale analysis feasible without a separate retrieval step. We propose three approaches: a single-agent approach that analyzes the entire code repository in one pass, and two multi-agent approaches based on self-verification and multi-agent debate. Our evaluation shows that the single-agent approach substantially improves recall, achieving 80.9%, a gain of 38.6 percentage points (pp) over the state-of-the-art retrieval-based baseline, albeit with a 25.8 pp reduction in precision. By leveraging collaborative reasoning, the multi-agent debate-based approach achieves the best overall precision-recall balance, with 24.5% precision and 71.0% recall, yielding an F2 score of 49.5%. This represents an improvement of 7.5 pp over the state of the art and 2.7 pp over the single-agent approach. While the quantitative evaluation establishes the overall effectiveness of the approaches, the aggregate metrics alone explain little due to the considerable variability across results. Importantly, qualitative analysis provides key insights into highly variable quantitative results: the retrieval approach suits requirements that echo the code's vocabulary; single-agent, full-context reasoning suits well-documented, consistently named code; and multi-agent debate suits indirect, conflicting, or distributed evidence.

## An Empirical Study on Clustering-based Quantum Approximate Optimization Algorithm for Regression Test Case Selection

DevFeed: [An Empirical Study on Clustering-based Quantum Approximate Optimization Algorithm for Regression Test Case Selection](<https://devfeed.tech/articles/an-empirical-study-on-clustering-based-quantum-approximate-optimization-algorithm-for-regression-test-case-selection-124041.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.09869>)

Author: Antonio Trovato, Martin Beseda, Dario Di Nucci

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [test](<https://devfeed.tech/topics/test.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>), [Combinatorial optimization](<https://devfeed.tech/topics/combinatorial-optimization.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [approach](<https://devfeed.tech/tags/approach.md>), [clustering](<https://devfeed.tech/tags/clustering.md>), [costs](<https://devfeed.tech/tags/costs.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [effectiveness](<https://devfeed.tech/tags/effectiveness.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [greedy](<https://devfeed.tech/tags/greedy.md>), [large](<https://devfeed.tech/tags/large.md>), [machines](<https://devfeed.tech/tags/machines.md>), [noise](<https://devfeed.tech/tags/noise.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [optimized](<https://devfeed.tech/tags/optimized.md>), [quantum](<https://devfeed.tech/tags/quantum.md>), [regression](<https://devfeed.tech/tags/regression.md>), [selection](<https://devfeed.tech/tags/selection.md>), [study](<https://devfeed.tech/tags/study.md>), [test](<https://devfeed.tech/tags/test.md>)

### AI overview

The study evaluates QAOA-TCS for regression test case selection, comparing it with classical and quantum approaches and testing robustness to quantum noise. It reports that QAOA-TCS outperforms state-of-the-art multi-objective approaches in most cases and remains stable under high noise; for single-objective formulations, QAOA-based methods perform comparably. SelectQA, based on quantum annealing, is reported as most efficient for large multi-objective problems, while QAOA-TCS offers the best effectiveness-efficiency trade-off for single-objective problems on realistic noisy simulators. The authors caution that comparisons are limited by the different computational environments used.

### Source excerpt

arXiv:2611.09869v1 Announce Type: new Abstract: Regression testing assesses whether software components that have already been developed and tested continue to work correctly after modifications. Ideally, regression testing re-executes the entire test suite after each modification, but this is infeasible in real-world scenarios because of the computational burden; therefore, optimized approaches, such as Test Case Selection, have been developed. The state of the art in regression testing optimization includes both classical approaches, such as Additional Greedy and DIV-GA, and quantum approaches, such as SelectQA, BootQA, and IGDec-QAOA. However, while the former incur high computational costs, the latter are constrained by the physical limitations of currently available quantum machines. This paper extends the evaluation of QAOA-TCS, our quantum approach for the test case selection problem. We thoroughly evaluate its performance, compare it with state-of-the-art classical and quantum approaches, and assess its robustness to quantum noise. The results show that QAOA-TCS outperforms state-of-the-art multi-objective approaches in most cases and maintains stable performance even under high noise levels. In the single-objective formulation, QAOA-based methods perform comparably. In terms of efficiency, our previous method based on quantum annealing, namely SelectQA, is the most efficient for large problems in the multi-objective formulation, whereas QAOA-TCS offers the best trade-off between effectiveness and efficiency in the single-objective formulation when run on realistic noisy simulators. However, such analysis is constrained by the different computational environments used to run the algorithms.

## KG-AST for Cheaper CI/CD: Budgeted Test Selection and AI Bug Review

DevFeed: [KG-AST for Cheaper CI/CD: Budgeted Test Selection and AI Bug Review](<https://devfeed.tech/articles/kg-ast-for-cheaper-ci-cd-budgeted-test-selection-and-ai-bug-review-120140.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.06950>)

Author: Henry Yeung

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [ai code review](<https://devfeed.tech/topics/ai-code-review.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [budget](<https://devfeed.tech/tags/budget.md>), [bug](<https://devfeed.tech/tags/bug.md>), [ci](<https://devfeed.tech/tags/ci.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [code-graph](<https://devfeed.tech/tags/code-graph.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [git](<https://devfeed.tech/tags/git.md>), [graph](<https://devfeed.tech/tags/graph.md>), [java](<https://devfeed.tech/tags/java.md>), [languages](<https://devfeed.tech/tags/languages.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [ranking](<https://devfeed.tech/tags/ranking.md>), [review](<https://devfeed.tech/tags/review.md>), [risk](<https://devfeed.tech/tags/risk.md>), [selection](<https://devfeed.tech/tags/selection.md>), [selector](<https://devfeed.tech/tags/selector.md>), [static](<https://devfeed.tech/tags/static.md>), [test](<https://devfeed.tech/tags/test.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

KG-AST prioritizes commits for heavy testing and AI review using git history and code-graph features. Across 155 projects in 11 languages, testing the riskiest 20% covered 49-64% of bug-introducing commits, versus 20% with random selection and 40-55% with largest-change-first selection. In Java, coverage was 64.6% versus 49.5% across 20 projects. The AI reviewer matched 26% of later-fixed defects from diffs alone and 32% with graph context; a static call-graph selector ran about 8% of a test suite while including a fix-written test in 85% of 67 cases. Labels were noisy, the budget metric assumes tests catch covered bugs, and no CI minutes were measured.

### Source excerpt

arXiv:2611.06950v1 Announce Type: cross Abstract: Continuous integration and AI-assisted review spend the same budget on every commit, although few commits introduce bugs. We study KG-AST, which scores each commit risk from git history and code-graph features before heavy tests or AI review run, so the budget goes to the riskiest commits first, and which hands the AI reviewer only a graph slice around the change. On 155 open-source projects in 11 languages, a single cross-language model trained on other projects ranks commits so that heavy testing of the riskiest 20% covers 49-64% of bug-introducing commits depending on the language, against 20% for random selection and 40-55% for testing the biggest change first. Every language beats the biggest-change baseline, by 3.8 to 13.2 points. In 20 Java projects the gain at a 20% budget is 15.1 points (64.6% vs 49.5%, 95% interval +11.6 to +18.4), and the model wins in 19 of the 20 projects. About twelve training projects suffice; richer features, tuning and cleaner labels did not help. An AI reviewer that sees only a commit's diff finds 26% of later-fixed defects exactly (32% with graph context), and about half of its findings that did not match a later fix were judged real by independent AI judges. A static call-graph selector runs about 8% of a test suite while including a test written for the fix in 85% of 67 fix commits. The Poincare embedding used in our retrieval work adds nothing measurable here: commit-risk AUC, bug review and severity ranking are unchanged or worse. The labels are noisy and the budget metric assumes a heavy test catches the bug it covers; no CI minutes were measured.

## Software Delivery Agent: From Code to Production

DevFeed: [Software Delivery Agent: From Code to Production](<https://devfeed.tech/articles/software-delivery-agent-from-code-to-production-119402.md>)

Original publisher: [Read original article](<https://www.harness.io/blog/beyond-the-coding-agent-how-the-software-delivery-agent-closes-the-gap-to-production>)

Author: Eric Minick

Published: 2026-11-04T00:00:00Z

Content type: article

Language: en

Sources: [Harness](<https://devfeed.tech/sources/harness-blog.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [CI/CD Pipeline](<https://devfeed.tech/topics/ci-cd-pipeline.md>), [Automation](<https://devfeed.tech/topics/automation.md>), [Continuous Delivery (CD)](<https://devfeed.tech/topics/continuous-delivery.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [blue green and canary deployments](<https://devfeed.tech/topics/blue-green-and-canary-deployments.md>), [Deployment Strategies](<https://devfeed.tech/topics/deployment-strategies.md>), [code security](<https://devfeed.tech/topics/code-security.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [blue-green deployment](<https://devfeed.tech/topics/blue-green-deployment.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding-assistants](<https://devfeed.tech/tags/ai-coding-assistants.md>), [ai-powered](<https://devfeed.tech/tags/ai-powered.md>), [automation](<https://devfeed.tech/tags/automation.md>), [blog](<https://devfeed.tech/tags/blog.md>), [code](<https://devfeed.tech/tags/code.md>), [coding](<https://devfeed.tech/tags/coding.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [gap](<https://devfeed.tech/tags/gap.md>), [harness](<https://devfeed.tech/tags/harness.md>), [production](<https://devfeed.tech/tags/production.md>), [reviews](<https://devfeed.tech/tags/reviews.md>), [security](<https://devfeed.tech/tags/security.md>), [software](<https://devfeed.tech/tags/software.md>), [software-delivery](<https://devfeed.tech/tags/software-delivery.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

Harness Software Delivery Agent targets the delivery work that follows code generation by combining shared organizational context, specialist reasoning agents, and worker agents that execute tasks under common permissions and an audit trail. Its Trust Gate evaluates whether changes pass by meeting requirements or by weakening tests and scans: the excerpt describes one clean change approved automatically, one test-gaming change blocked, and one legitimate test removal that still required human sign-off under policy. Other described controls include governed pipeline templates, policy checks that prevent removal of required production approvals, canary rollbacks based on error rates, selective builds and tests, and security finding deduplication with AI-generated fix pull requests. The text reports customer results including Citi reducing build-to-production release time to about seven minutes, United Airlines deployments 75% faster, and Ancestry reporting lower onboarding effort, fewer deployment-related outages, and higher deployment velocity. These are claims in the supplied excerpt; it does not provide methodology or full coverage of the webinar.

### Source excerpt

Discover how Harness Software Delivery Agent closes the gap between coding and production with AI-powered reviews, testing, security, and deployment automation. | Blog

## Work in progress is the new software delivery bottleneck

DevFeed: [Work in progress is the new software delivery bottleneck](<https://devfeed.tech/articles/work-in-progress-is-the-new-software-delivery-bottleneck-109925.md>)

Original publisher: [Read original article](<https://circleci.com/blog/work-in-progress-delivery-bottleneck/>)

Author: Jacob Schmitt

Published: 2026-10-30T16:00:00Z

Content type: article

Language: en

Sources: [CircleCI](<https://devfeed.tech/sources/the-circleci-blog-feed-circleci.md>)

Topics: [software delivery metrics](<https://devfeed.tech/topics/software-delivery-metrics.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [merge efficiency ratio](<https://devfeed.tech/topics/merge-efficiency-ratio.md>), [Kanban](<https://devfeed.tech/topics/kanban.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [engineering-leadership](<https://devfeed.tech/topics/engineering-leadership.md>)

Tags: [1-7](<https://devfeed.tech/tags/1-7.md>), [2](<https://devfeed.tech/tags/2.md>), [2026](<https://devfeed.tech/tags/2026.md>), [3](<https://devfeed.tech/tags/3.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [but](<https://devfeed.tech/tags/but.md>), [circleci-news](<https://devfeed.tech/tags/circleci-news.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [developer-productivity](<https://devfeed.tech/tags/developer-productivity.md>), [flat](<https://devfeed.tech/tags/flat.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [little-s-law](<https://devfeed.tech/tags/little-s-law.md>), [median](<https://devfeed.tech/tags/median.md>), [parallel-development](<https://devfeed.tech/tags/parallel-development.md>), [popular](<https://devfeed.tech/tags/popular.md>), [progress](<https://devfeed.tech/tags/progress.md>), [software](<https://devfeed.tech/tags/software.md>), [software-delivery](<https://devfeed.tech/tags/software-delivery.md>), [software-delivery-bottlenecks](<https://devfeed.tech/tags/software-delivery-bottlenecks.md>), [teams](<https://devfeed.tech/tags/teams.md>), [throughput](<https://devfeed.tech/tags/throughput.md>), [wip-in-software-development](<https://devfeed.tech/tags/wip-in-software-development.md>), [wip-limits](<https://devfeed.tech/tags/wip-limits.md>), [with](<https://devfeed.tech/tags/with.md>), [work](<https://devfeed.tech/tags/work.md>), [work-in-progress](<https://devfeed.tech/tags/work-in-progress.md>), [work-in-progress-limits](<https://devfeed.tech/tags/work-in-progress-limits.md>)

### AI overview

Coding agents let teams pursue more work in parallel, but median main-branch throughput stayed at 1.7 runs per day through 2026 even as active projects and branches surged. The article argues that growing work in progress (WIP) can overload validation, review, and integration; under Little's Law, cycle time rises when WIP grows without higher throughput. The top 5% of organizations by main-branch activity increased active feature branches by 55% and throughput by 31%, showing that parallel work can scale alongside delivery. Suggested practices are to track branches, open PRs, blocked work, and review queues; set practical WIP limits and clear agent assignments; provide agents with fast, actionable CI feedback; and measure cycle time, first-review time, merges to main, and merge efficiency ratio rather than activity alone. CircleCI tools named include Chunk sidecars, Smarter Testing, and its CLI and MCP server for validation feedback and failure diagnostics; developers still handle decisions requiring engineering judgment.

### Source excerpt

Teams are opening more branches with coding agents, but median throughput on main is flat. Learn why WIP is the new delivery bottleneck and how to fix it.

## CI Test Suite Slow? 6 Fixes We Measured in Harness

DevFeed: [CI Test Suite Slow? 6 Fixes We Measured in Harness](<https://devfeed.tech/articles/ci-test-suite-slow-6-fixes-we-measured-in-harness-130792.md>)

Original publisher: [Read original article](<https://www.harness.io/blog/flaky-tests-slow-ci-pipeline>)

Author: Shibam Dhar

Published: 2026-10-29T01:00:00Z

Content type: article

Language: en

Sources: [Harness](<https://devfeed.tech/sources/harness-blog.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [dynamic test splitting](<https://devfeed.tech/topics/dynamic-test-splitting.md>), [intermittent failures](<https://devfeed.tech/topics/intermittent-failures.md>), [Continuous integration](<https://devfeed.tech/topics/continuous-integration.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [account](<https://devfeed.tech/tags/account.md>), [added](<https://devfeed.tech/tags/added.md>), [after](<https://devfeed.tech/tags/after.md>), [ci](<https://devfeed.tech/tags/ci.md>), [flaky](<https://devfeed.tech/tags/flaky.md>), [harness](<https://devfeed.tech/tags/harness.md>), [intelligence](<https://devfeed.tech/tags/intelligence.md>), [isolation](<https://devfeed.tech/tags/isolation.md>), [lab](<https://devfeed.tech/tags/lab.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [performance](<https://devfeed.tech/tags/performance.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [sharding](<https://devfeed.tech/tags/sharding.md>), [state](<https://devfeed.tech/tags/state.md>), [test](<https://devfeed.tech/tags/test.md>), [time](<https://devfeed.tech/tags/time.md>), [vs](<https://devfeed.tech/tags/vs.md>)

### AI overview

In a controlled Harness lab, the full pipeline took about 100 seconds while Vitest ran in about 2 seconds, with initialization varying from 6 to 42 seconds. Measure pipeline steps before optimizing the test runner. Retries can make flaky tests appear green; the Harness Test Analysis Plugin can quarantine named failures while still rejecting failures absent from the quarantine list. Its demonstration used deliberate failures and did not test automatic flake detection. Isolate mutable test state before adding concurrency: fresh state changed four tests from two failures to four passes. Two shards did not speed up this small suite because initialization dominated. npm caching worked after using `/harness/.npm`, but restore took eight seconds and the warm install saved one second, so measure the full cost. In a separate Java/JUnit fixture, Harness Test Intelligence selected 17 of 21 tests after a change to DiscountEngine.java, skipping four unrelated tax tests; the time saving on this small suite was negligible. The lab recommends fixing determinism first, then measuring whether caching, sharding, or change-aware selection pays off. These results are from controlled examples, not a way

### Source excerpt

CI performance lab measuring test time vs pipeline time, flaky test quarantine, shared-state isolation, parallel sharding and Harness Test Intelligence. | Blog

## Vercel CI is now in beta

DevFeed: [Vercel CI is now in beta](<https://devfeed.tech/articles/vercel-ci-is-now-in-beta-105508.md>)

Original publisher: [Read original article](<https://vercel.com/changelog/vercel-ci-is-now-in-beta>)

Author: Cody Wong

Published: 2026-10-28T19:00:00Z

Content type: release

Language: en

Sources: [Vercel News](<https://devfeed.tech/sources/vercel-news.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Vercel](<https://devfeed.tech/topics/vercel.md>), [Pull Request](<https://devfeed.tech/topics/pull-request.md>)

Tags: [beta](<https://devfeed.tech/tags/beta.md>), [caching](<https://devfeed.tech/tags/caching.md>), [ci](<https://devfeed.tech/tags/ci.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [configuration](<https://devfeed.tech/tags/configuration.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [git](<https://devfeed.tech/tags/git.md>), [lint](<https://devfeed.tech/tags/lint.md>), [turborepo](<https://devfeed.tech/tags/turborepo.md>), [vercel](<https://devfeed.tech/tags/vercel.md>)

### AI overview

Vercel CI is in beta and can be enabled for a Turborepo with a `ci` tag on tasks; tagged tasks run on every push. The excerpt says the service runs only CI affected by code changes, uses Vercel Remote Cache for task caching, installs dependencies once per run, and caches dependencies across code updates. Setup involves running `npx skills use vercel/ci@setup`, selecting initial tasks, configuring CI, and opening a pull request. Vercel reports 2-6x shorter time from push to "CI passed" in its internal repositories; this is an internal result, not a general performance guarantee.

### Source excerpt

You can now run CI for your Turborepo with one line of configuration. In Vercel's internal repositories, our wait time from git push to "CI passed" is 2-6x shorter, dramatically improving our CI's ability to keep up with our coding agents. To get started, prompt your agent in your repo: Each task marked with a ci tag runs on every push. Vercel CI automatically takes care of: Only running CI that is affected by your code changes Task caching with Vercel Remote Cache Installing dependencies once per run Caching dependencies across code updates To learn more: Check out our blog post Visit the documentation Discuss Vercel CI on the Community announcement Read more

## Introducing Vercel CI

DevFeed: [Introducing Vercel CI](<https://devfeed.tech/articles/introducing-vercel-ci-105504.md>)

Original publisher: [Read original article](<https://vercel.com/blog/introducing-vercel-ci>)

Author: Cody Wong

Published: 2026-10-28T07:00:00Z

Content type: release

Language: en

Sources: [Vercel News](<https://devfeed.tech/sources/vercel-news.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Vercel](<https://devfeed.tech/topics/vercel.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [branch protection rules](<https://devfeed.tech/topics/branch-protection-rules.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [beta](<https://devfeed.tech/tags/beta.md>), [branch-protection-rules](<https://devfeed.tech/tags/branch-protection-rules.md>), [cache](<https://devfeed.tech/tags/cache.md>), [ci](<https://devfeed.tech/tags/ci.md>), [cli](<https://devfeed.tech/tags/cli.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [compute](<https://devfeed.tech/tags/compute.md>), [dashboard](<https://devfeed.tech/tags/dashboard.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [documentation](<https://devfeed.tech/tags/documentation.md>), [failure](<https://devfeed.tech/tags/failure.md>), [faster](<https://devfeed.tech/tags/faster.md>), [flaky](<https://devfeed.tech/tags/flaky.md>), [git](<https://devfeed.tech/tags/git.md>), [github](<https://devfeed.tech/tags/github.md>), [hive](<https://devfeed.tech/tags/hive.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [merge](<https://devfeed.tech/tags/merge.md>), [vercel](<https://devfeed.tech/tags/vercel.md>)

### AI overview

Vercel CI Beta uses Turborepo's task graph and change detection to run only affected tasks, reuse cached outputs for unchanged inputs, and start each remaining task on its own Hive-powered machine after dependencies finish. Existing Turborepo projects can use their `turbo.json`; other repositories can add that file and the `turbo` package, including single-package repos, then tag tasks with `"ci"`. Remote Cache is enabled by default, with no separate workflow YAML. The article reports Vercel's push-to-pass time as 2-6x faster, but gives no benchmark method or broader performance guarantee. Runs can be inspected in the Vercel dashboard, CLI, or MCP; GitHub checks can work with existing branch protection. Beta is available on all plans. Planned additions include advanced observability, distributed execution before pushing, and more granular test selection.

### Source excerpt

Coding agents are opening more pull requests than ever, and each one waits on CI before anyone can review or merge it. Most CI systems run the entire workflow on every push, even a one-line change. The more your agents push, the more time you spend waiting on CI. Today, we're announcing Vercel CI in Beta. It runs only the tasks affected by your code change, and spins up only the compute those checks need, the same way Fluid compute scales with your traffic. At Vercel, our CI now keeps up with our coding agents. Time from git push to "CI passed" is 2-6x faster than it was before. How it works Vercel CI uses Turborepo to understand your tasks and which ones depend on each other. It reads turbo.json, so if you're already using Turborepo, there's no separate CI config to write. If you're not on Turborepo yet, add turbo.json and the turbo package. It works in single-package repos too. Then tag the tasks you want Vercel CI to run. "tags": ["ci"] is all you need to get started. No workflow YAML to maintain, no drift between your CI config and turbo.json, and Remote Cache is on by default. When you or your agent push code, Vercel CI determines which tasks need to run based on the change. Tasks whose inputs haven't changed come from cache. The remaining tasks each run on their own machine, powered by Hive, Vercel's compute layer, and start as soon as their dependencies finish. Vercel CI gives each task the compute it needs and nothing more. Running your CI on Vercel To get started, follow the documentation or prompt your agent from your repo: Once set up, you or your agent can watch and inspect CI through the Vercel dashboard, Vercel CLI, or Vercel MCP. When something fails, go straight to the failing task, see what's blocked behind it, and open its logs, or search across recent runs to see whether a test has been flaky all week. Vercel CI posts a check to your pull request on GitHub, so your existing branch protection rules can block merges on failure, the same way they do

## AI Testing Strategy for SaaS Teams

DevFeed: [AI Testing Strategy for SaaS Teams](<https://devfeed.tech/articles/ai-testing-strategy-for-saas-teams-102331.md>)

Original publisher: [Read original article](<https://hackernoon.com/ai-testing-strategy-for-saas-teams?source=rss>)

Author: QA.tech

Published: 2026-10-27T12:30:25Z

Content type: article

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [Test automation](<https://devfeed.tech/topics/test-automation.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [ui-testing](<https://devfeed.tech/topics/ui-testing.md>), [Code review](<https://devfeed.tech/topics/code-review.md>), [staging](<https://devfeed.tech/topics/staging.md>), [AI in frontend development](<https://devfeed.tech/topics/ai-in-frontend-development.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-powered](<https://devfeed.tech/tags/ai-powered.md>), [ai-testing](<https://devfeed.tech/tags/ai-testing.md>), [authentication](<https://devfeed.tech/tags/authentication.md>), [billing](<https://devfeed.tech/tags/billing.md>), [blocking](<https://devfeed.tech/tags/blocking.md>), [bug](<https://devfeed.tech/tags/bug.md>), [faster](<https://devfeed.tech/tags/faster.md>), [good-company](<https://devfeed.tech/tags/good-company.md>), [pr](<https://devfeed.tech/tags/pr.md>), [quality-assurance](<https://devfeed.tech/tags/quality-assurance.md>), [regression](<https://devfeed.tech/tags/regression.md>), [regression-testing](<https://devfeed.tech/tags/regression-testing.md>), [release](<https://devfeed.tech/tags/release.md>), [risk](<https://devfeed.tech/tags/risk.md>), [saas](<https://devfeed.tech/tags/saas.md>), [saas-testing](<https://devfeed.tech/tags/saas-testing.md>), [scheduled](<https://devfeed.tech/tags/scheduled.md>), [ship](<https://devfeed.tech/tags/ship.md>), [software-testing](<https://devfeed.tech/tags/software-testing.md>), [strategy](<https://devfeed.tech/tags/strategy.md>), [test-automation](<https://devfeed.tech/tags/test-automation.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

Place tests at the stage where they can guide release decisions: block each PR on risk-weighted checks for affected flows and critical paths such as authentication and billing, run a small curated insurance suite on a schedule, and use human or agent exploratory testing for subjective or novel cases. When preview environments are unavailable, post-merge staging tests offer similar coverage but cannot block the merge and provide less clear attribution. Exhaustive SaaS state-matrix coverage is impractical, and sterile seeded data or extensive mocking can hide production issues. The excerpt describes QA.tech as running dynamic PR tests and scheduled checks, with auditable verdict evidence; it also notes those checks do not cover every state combination.

### Source excerpt

Learn how SaaS teams can use AI-powered testing, risk-based PR checks, and scheduled regression suites to ship faster with greater release con

## How to run fewer tests and speed up CI with CircleCI Smarter Testing

DevFeed: [How to run fewer tests and speed up CI with CircleCI Smarter Testing](<https://devfeed.tech/articles/how-to-run-fewer-tests-and-speed-up-ci-with-circleci-smarter-testing-96424.md>)

Original publisher: [Read original article](<https://circleci.com/blog/run-fewer-tests-speed-up-ci-smarter-testing/>)

Author: Roger Winter

Published: 2026-10-22T19:00:00Z

Content type: tutorial

Language: en

Sources: [CircleCI](<https://devfeed.tech/sources/the-circleci-blog-feed-circleci.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [auto rerun failed tests](<https://devfeed.tech/topics/auto-rerun-failed-tests.md>), [CircleCI](<https://devfeed.tech/topics/circleci.md>), [dynamic test splitting](<https://devfeed.tech/topics/dynamic-test-splitting.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Integration testing](<https://devfeed.tech/topics/integration-testing.md>)

Tags: [agent-skills](<https://devfeed.tech/tags/agent-skills.md>), [api](<https://devfeed.tech/tags/api.md>), [auto-rerun-failed-tests](<https://devfeed.tech/tags/auto-rerun-failed-tests.md>), [ci](<https://devfeed.tech/tags/ci.md>), [circleci](<https://devfeed.tech/tags/circleci.md>), [circleci-cli](<https://devfeed.tech/tags/circleci-cli.md>), [circleci-news](<https://devfeed.tech/tags/circleci-news.md>), [circleci-testsuite](<https://devfeed.tech/tags/circleci-testsuite.md>), [cli](<https://devfeed.tech/tags/cli.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [compute-costs](<https://devfeed.tech/tags/compute-costs.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cypress](<https://devfeed.tech/tags/cypress.md>), [dynamic-test-splitting](<https://devfeed.tech/tags/dynamic-test-splitting.md>), [engineering-productivity](<https://devfeed.tech/tags/engineering-productivity.md>), [flaky](<https://devfeed.tech/tags/flaky.md>), [flaky-tests](<https://devfeed.tech/tags/flaky-tests.md>), [retry](<https://devfeed.tech/tags/retry.md>), [smarter-testing](<https://devfeed.tech/tags/smarter-testing.md>), [test-impact-analysis](<https://devfeed.tech/tags/test-impact-analysis.md>), [testing](<https://devfeed.tech/tags/testing.md>), [tests](<https://devfeed.tech/tags/tests.md>), [typescript](<https://devfeed.tech/tags/typescript.md>)

### AI overview

CircleCI Smarter Testing uses test coverage to run tests affected by a change, retries failed test files within the same job, and dynamically splits work across parallel nodes. Setup requires test suites with discovery and run commands, JUnit reports containing test file paths, and-- for impact analysis--a runner coverage plugin plus full-test-run paths for shared configuration and setup files. In the TypeScript/Jest example, one-file changes reduced the median unit test step from 16.4 to 6.4 seconds across 12 paired runs; mean executor time per pull request fell from 120 to 84 seconds. Selection still caught an integration regression, while a separate demonstration reports 44 of 9,716 tests selected and runtime dropping from 3:56 to 22 seconds. Impact analysis and auto rerun are paid features counted by discovered test atoms; static timing-based splitting through the CLI is free. Keep auto rerun limits low and fix pass-on-retry failures, since real race conditions can appear flaky. Coverage only reflects code executed in tests, so the example retains a type check or build on every commit. Figures are specific to the described examples and benchmarks.

### Source excerpt

Set up CircleCI Smarter Testing on a TypeScript app: run only the tests a change affects, auto-retry flaky tests, and measure the CI time you save.

## mabl Introduces Agentic Testing Harness to Clear the Verification Bottleneck Created by Coding Agents

DevFeed: [mabl Introduces Agentic Testing Harness to Clear the Verification Bottleneck Created by Coding Agents](<https://devfeed.tech/articles/mabl-introduces-agentic-testing-harness-to-clear-the-verification-bottleneck-created-by-coding-agents-96150.md>)

Original publisher: [Read original article](<https://sdtimes.com/ai-testing/mabl-introduces-agentic-testing-harness-to-clear-the-verification-bottleneck-created-by-coding-agents/>)

Author: SD Times Newswire

Published: 2026-10-22T16:31:30Z

Content type: release

Language: en

Sources: [SD Times](<https://devfeed.tech/sources/sd-times.md>)

Topics: [Test automation](<https://devfeed.tech/topics/test-automation.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>), [Azure DevOps](<https://devfeed.tech/topics/azure-devops.md>)

Tags: [agent-guardrails](<https://devfeed.tech/tags/agent-guardrails.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-testing](<https://devfeed.tech/tags/agentic-testing.md>), [agents](<https://devfeed.tech/tags/agents.md>), [boston](<https://devfeed.tech/tags/boston.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [change](<https://devfeed.tech/tags/change.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [guardrails](<https://devfeed.tech/tags/guardrails.md>), [harness](<https://devfeed.tech/tags/harness.md>), [impact](<https://devfeed.tech/tags/impact.md>), [independent](<https://devfeed.tech/tags/independent.md>), [introduced](<https://devfeed.tech/tags/introduced.md>), [knowledge-graph](<https://devfeed.tech/tags/knowledge-graph.md>), [latest-news](<https://devfeed.tech/tags/latest-news.md>), [newswire](<https://devfeed.tech/tags/newswire.md>), [platform](<https://devfeed.tech/tags/platform.md>), [pull-request](<https://devfeed.tech/tags/pull-request.md>), [quality](<https://devfeed.tech/tags/quality.md>), [release](<https://devfeed.tech/tags/release.md>), [test-case-management](<https://devfeed.tech/tags/test-case-management.md>), [testing](<https://devfeed.tech/tags/testing.md>), [verification](<https://devfeed.tech/tags/verification.md>)

### AI overview

mabl introduced a testing harness that gives engineering and quality teams an independent check on software produced by coding agents. Teams can define agent objectives and guardrails; mabl's knowledge graph assesses change impact, identifies relevant tests and coverage gaps; and developers can view results from the terminal before opening a pull request. The release also includes cross-workspace test suite coordination and integrations that use cases in Xray, Zephyr, TestRail and Azure DevOps as standards, with results returned to those systems. Desktop application testing is planned for enterprise customers and is coming soon. mabl reports that in the prior 90 days its agents analyzed 1.58 million unique failures across 3.3 million sessions, made 3.0 million MCP calls and created over 4,200 tests; it estimates this saved customers about 100,000 engineering hours.

### Source excerpt

BOSTON -- mabl, the agentic testing platform, today introduced new capabilities that give engineering and quality teams independent verification of the software their coding agents produce. Teams can now set the standards and guardrails their agents are held to and see the impact of a change from the terminal before a pull request. The release... ... continue reading The post mabl Introduces Agentic Testing Harness to Clear the Verification Bottleneck Created by Coding Agents appeared first on SD Times.

## Stacked PRs: what they are and how to run CI on a stack

DevFeed: [Stacked PRs: what they are and how to run CI on a stack](<https://devfeed.tech/articles/stacked-prs-what-they-are-and-how-to-run-ci-on-a-stack-88989.md>)

Original publisher: [Read original article](<https://circleci.com/blog/stacked-prs-and-ci/>)

Author: Jacob Schmitt

Published: 2026-10-16T16:00:00Z

Content type: tutorial

Language: en

Sources: [CircleCI](<https://devfeed.tech/sources/the-circleci-blog-feed-circleci.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [dynamic test splitting](<https://devfeed.tech/topics/dynamic-test-splitting.md>), [developer-productivity](<https://devfeed.tech/topics/developer-productivity.md>), [auto rerun failed tests](<https://devfeed.tech/topics/auto-rerun-failed-tests.md>), [CircleCI](<https://devfeed.tech/topics/circleci.md>)

Tags: [circleci](<https://devfeed.tech/tags/circleci.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [continuous-integration](<https://devfeed.tech/tags/continuous-integration.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [engineering-productivity](<https://devfeed.tech/tags/engineering-productivity.md>), [github-stacked-prs](<https://devfeed.tech/tags/github-stacked-prs.md>), [path-filtering](<https://devfeed.tech/tags/path-filtering.md>), [pull-requests](<https://devfeed.tech/tags/pull-requests.md>), [smarter-testing](<https://devfeed.tech/tags/smarter-testing.md>), [stacked-diffs](<https://devfeed.tech/tags/stacked-diffs.md>), [stacked-prs](<https://devfeed.tech/tags/stacked-prs.md>), [stacked-prs-ci](<https://devfeed.tech/tags/stacked-prs-ci.md>), [stacked-pull-requests](<https://devfeed.tech/tags/stacked-pull-requests.md>), [test-impact-analysis](<https://devfeed.tech/tags/test-impact-analysis.md>), [tests](<https://devfeed.tech/tags/tests.md>), [what-is-a-stacked-pr](<https://devfeed.tech/tags/what-is-a-stacked-pr.md>)

### AI overview

This tutorial explains how stacked pull requests work and why each branch can trigger repeated CI pipelines. It describes ways to reduce redundant validation, including path filtering, test impact analysis, dynamic test splitting and parallelism, pre-push checks with coding-agent sidecars, and tools for diagnosing CI failures. It also explains tradeoffs such as coverage across stacked changes, compute costs, and when a stack may add more review work than it saves.

### Source excerpt

Learn how stacked pull requests work, why they increase CI usage, and how to reduce redundant builds and tests with CircleCI.

## Running a Green CI Pipeline Isn't the Hard Part Anymore

DevFeed: [Running a Green CI Pipeline Isn't the Hard Part Anymore](<https://devfeed.tech/articles/running-a-green-ci-pipeline-isn-t-the-hard-part-anymore-88369.md>)

Original publisher: [Read original article](<https://www.harness.io/blog/green-ci-pipeline-isnt-enough>)

Author: Shibam Dhar

Published: 2026-10-16T09:58:51.179334Z

Content type: tutorial

Language: en

Sources: [Harness](<https://devfeed.tech/sources/harness-blog.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [dynamic test splitting](<https://devfeed.tech/topics/dynamic-test-splitting.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [Build tool](<https://devfeed.tech/topics/build-tool.md>), [Deployment Strategies](<https://devfeed.tech/topics/deployment-strategies.md>)

Tags: [ai-assisted-coding](<https://devfeed.tech/tags/ai-assisted-coding.md>), [build-times](<https://devfeed.tech/tags/build-times.md>), [caching](<https://devfeed.tech/tags/caching.md>), [speed](<https://devfeed.tech/tags/speed.md>), [test](<https://devfeed.tech/tags/test.md>)

### AI overview

This tutorial describes how Harness CI uses Cache Intelligence, Build Intelligence, and Test Intelligence to reduce redundant work in AI-driven CI pipelines. In a Java Maven example, Test Intelligence selected 12 of 42 tests after a single-file change, while other features reuse dependency caches and compiled outputs.

### Source excerpt

AI-assisted coding is changing CI pipelines. Learn how Harness CI uses Cache, Build and Test Intelligence to reduce redundant work and speed up builds. | Blog

## Making testing work better for AI driven development

DevFeed: [Making testing work better for AI driven development](<https://devfeed.tech/articles/making-testing-work-better-for-ai-driven-development-20426.md>)

Original publisher: [Read original article](<https://semaphore.io/blog/making-testing-work-better-for-ai-driven-development>)

Author: Pete Miloravac

Published: 2026-09-10T09:28:14Z

Content type: release

Language: en

Sources: [Semaphore Engineering](<https://devfeed.tech/sources/semaphore-engineering.md>)

Topics: [Software Testing](<https://devfeed.tech/topics/software-testing.md>), [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>), [software-development](<https://devfeed.tech/topics/software-development.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [test-coverage](<https://devfeed.tech/topics/test-coverage.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [context](<https://devfeed.tech/tags/context.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [detection](<https://devfeed.tech/tags/detection.md>), [failure](<https://devfeed.tech/tags/failure.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [flaky-tests](<https://devfeed.tech/tags/flaky-tests.md>), [framework](<https://devfeed.tech/tags/framework.md>), [frameworks](<https://devfeed.tech/tags/frameworks.md>), [graphs](<https://devfeed.tech/tags/graphs.md>), [integrations](<https://devfeed.tech/tags/integrations.md>), [making](<https://devfeed.tech/tags/making.md>), [native](<https://devfeed.tech/tags/native.md>), [notifications](<https://devfeed.tech/tags/notifications.md>), [product-news](<https://devfeed.tech/tags/product-news.md>), [product-update](<https://devfeed.tech/tags/product-update.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

Semaphore is exploring ways to give developers and AI agents more useful test feedback: collect richer, framework-specific results (with initial support for xUnit and RSpec), detect flaky tests using weighted signals such as reruns and inconsistent results for the same commit, and select relevant tests based on code graphs. Flakiness scores would fall over time when results stabilize, with threshold alerts available through the product, notifications, or webhooks. Code-graph test selection remains early research; initial results across some technology stacks are described as promising. The excerpt frames these efforts as ways to improve failure diagnosis and shorten feedback loops while preserving relevant test coverage.

### Source excerpt

AI agents can generate and modify code quickly. But when a test fails, speed is no longer enough. The agent needs enough context to understand what happened, why it happened, and whether the code or the test itself needs to change. In our latest product update, Marko Gaćeša is joined by Semaphore engineers Nick and [...] The post Making testing work better for AI driven development appeared first on Semaphore.

## Supercharging Continuous Integration with Gradle

DevFeed: [Supercharging Continuous Integration with Gradle](<https://devfeed.tech/articles/supercharging-continuous-integration-with-gradle-15901.md>)

Original publisher: [Read original article](<https://developer.squareup.com/blog/supercharging-continuous-integration-with-gradle>)

Author: Paul Hundal

Published: 2023-07-27T14:00:00Z

Content type: article

Language: en

Sources: [Square Corner](<https://devfeed.tech/sources/square-corner-blog-medium.md>)

Topics: [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Gradle](<https://devfeed.tech/topics/gradle.md>), [Continuous integration](<https://devfeed.tech/topics/continuous-integration.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [bazel](<https://devfeed.tech/topics/bazel.md>), [Benchmark](<https://devfeed.tech/topics/benchmark.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [22](<https://devfeed.tech/tags/22.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [addition](<https://devfeed.tech/tags/addition.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [all](<https://devfeed.tech/tags/all.md>), [allowed](<https://devfeed.tech/tags/allowed.md>), [also](<https://devfeed.tech/tags/also.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [android](<https://devfeed.tech/tags/android.md>), [api](<https://devfeed.tech/tags/api.md>), [bazel](<https://devfeed.tech/tags/bazel.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [continuous-integration](<https://devfeed.tech/tags/continuous-integration.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [gradle](<https://devfeed.tech/tags/gradle.md>), [integration](<https://devfeed.tech/tags/integration.md>), [journey](<https://devfeed.tech/tags/journey.md>), [like](<https://devfeed.tech/tags/like.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

Square's Android CI project grew beyond 5,000 Gradle modules, with builds averaging 65 minutes. Kochiku distributes work across 1,200 shards; shard avoidance skips unaffected shards, averaging 600 skipped per build. To identify affected work, Square moved from regex-based checks and filesystem YAML metadata to Affected Paths, which queries Gradle's daemon through the Tooling API and reduced analysis time and memory use. An Avoidance Cache stores Gradle dependency graph responses in S3 keyed by build-file hashes; because only 8% of builds changed build files, the expected cache hit rate was 92%, and hits can bypass Gradle. Avoidance analysis fell from 11 minutes to 2 minutes, an 80% reduction. Static AST parsing could reduce Gradle configuration work, but missed transitive dependencies created by dynamic build logic. Sharing Gradle's Configuration Cache across CI remained under investigation: the Tooling API was not compatible with it at the time described.

### Source excerpt

Unveiling the journey of shard avoidance and Bazel-like queryability

## Incremental CodeQL

DevFeed: [Incremental CodeQL](<https://devfeed.tech/articles/incremental-codeql-67161.md>)

Original publisher: [Read original article](<https://githubnext.com/projects/incremental-codeql/>)

Published: 2022-10-11T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Static code analysis](<https://devfeed.tech/topics/static-code-analysis.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>)

Tags: [analysis](<https://devfeed.tech/tags/analysis.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [codeql](<https://devfeed.tech/tags/codeql.md>), [incremental](<https://devfeed.tech/tags/incremental.md>), [performance](<https://devfeed.tech/tags/performance.md>)

### AI overview

The Incremental CodeQL project investigates reusing prior static analysis results so code scanning can update findings in proportion to code changes. Experiments with a prototype evaluator found updates typically took a few seconds and stayed below 15 seconds even for the largest commits, while initial analysis could take an hour and use tens of gigabytes of memory. The results support further work, but memory use remains a concern before production deployment.

### Source excerpt

The goal of the Incremental CodeQL project is to investigate the potential for incrementalization to improve the performance of CodeQL query evaluation. Instead of having to re-analyze the whole code base from scratch, incremental evaluation reuses the previously computed analysis results and only re-analyzes the parts of the code base that are affected by a program change. This approach can lead to significant performance speed-ups without the need to make any changes to the analysis implementation.

## April service disruptions analysis

DevFeed: [April service disruptions analysis](<https://devfeed.tech/articles/april-service-disruptions-analysis-68849.md>)

Original publisher: [Read original article](<https://github.blog/news-insights/company-news/april-service-disruptions-analysis/>)

Author: Keith Ballinger

Published: 2020-05-22T16:00:01Z

Content type: article

Language: en

Sources: [GitHub Blog](<https://devfeed.tech/sources/github-engineering.md>)

Topics: [reliability-engineering](<https://devfeed.tech/topics/reliability-engineering.md>), [incident management](<https://devfeed.tech/topics/incident-management.md>), [zero downtime deployment](<https://devfeed.tech/topics/zero-downtime-deployment.md>), [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>)

Tags: [bare-metal](<https://devfeed.tech/tags/bare-metal.md>), [bgp](<https://devfeed.tech/tags/bgp.md>), [canary](<https://devfeed.tech/tags/canary.md>), [ci](<https://devfeed.tech/tags/ci.md>), [company-news](<https://devfeed.tech/tags/company-news.md>), [database](<https://devfeed.tech/tags/database.md>), [developers](<https://devfeed.tech/tags/developers.md>), [news-insights](<https://devfeed.tech/tags/news-insights.md>), [reliability](<https://devfeed.tech/tags/reliability.md>), [service-interruptions](<https://devfeed.tech/tags/service-interruptions.md>)

### AI overview

GitHub reports three April service incidents that caused sitewide unavailability or degraded service for 5 hours and 36 minutes. The causes included load balancer and database connection misconfigurations and a network routing change. GitHub describes work to improve staging and canary coverage, add network configuration testing through CI, and better match staging database setups to production.

### Source excerpt

We had multiple service interruptions in April that may have impacted your projects and businesses. We know how important reliability is for our users and have detailed an analysis on the disruptions.

## Experiment: Adjusting SETA to run individual files instead of individual jobs

DevFeed: [Experiment: Adjusting SETA to run individual files instead of individual jobs](<https://devfeed.tech/articles/experiment-adjusting-seta-to-run-individual-files-instead-of-individual-jobs-20259.md>)

Original publisher: [Read original article](<https://elvis314.wordpress.com/2018/08/27/experiment-adjusting-seta-to-run-individual-files-instead-of-individual-jobs/>)

Author: elvis314

Published: 2018-08-27T18:17:18Z

Content type: article

Language: en

Sources: [Mozilla Automation](<https://devfeed.tech/sources/mozilla-automation.md>)

Topics: [test impact analysis](<https://devfeed.tech/topics/test-impact-analysis.md>), [Unit testing](<https://devfeed.tech/topics/unit-testing.md>), [Developer Tools](<https://devfeed.tech/topics/developer-tools.md>)

Tags: [aggregation](<https://devfeed.tech/tags/aggregation.md>), [android](<https://devfeed.tech/tags/android.md>), [ci](<https://devfeed.tech/tags/ci.md>), [code-coverage](<https://devfeed.tech/tags/code-coverage.md>), [configs](<https://devfeed.tech/tags/configs.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cpu-time](<https://devfeed.tech/tags/cpu-time.md>), [experiment](<https://devfeed.tech/tags/experiment.md>), [failed](<https://devfeed.tech/tags/failed.md>), [files](<https://devfeed.tech/tags/files.md>), [future](<https://devfeed.tech/tags/future.md>), [harness](<https://devfeed.tech/tags/harness.md>), [jobs](<https://devfeed.tech/tags/jobs.md>), [leaks](<https://devfeed.tech/tags/leaks.md>), [manifest](<https://devfeed.tech/tags/manifest.md>), [minutes](<https://devfeed.tech/tags/minutes.md>), [multiple](<https://devfeed.tech/tags/multiple.md>), [osx](<https://devfeed.tech/tags/osx.md>), [project](<https://devfeed.tech/tags/project.md>), [queries](<https://devfeed.tech/tags/queries.md>), [uncategorized](<https://devfeed.tech/tags/uncategorized.md>)

### AI overview

SETA reduces test load by prioritizing jobs that have found regressions, but those jobs may rerun many tests when only one failed, and chunking can make failures harder to target. In a June experiment, queries of Treeherder data identified 171 manifest-listed tests, run in six jobs using 147 CPU minutes. The author sees potential in selecting high-priority tests with SETA and adding tests identified through code coverage. The approach covered about 75% of failures because some issues, such as hangs or tooling changes, cannot be tied to a manifest test. It also lacked load balancing across configurations: running the tests on 27 configs could erase the savings. Extracting unique failures required substantial database querying and changes to SETA's aggregation code.

### Source excerpt

3.5 years ago we implemented and integrated SETA. This has a net effect today of reducing our load between 60-70%. SETA works on the premise of identifying specific test jobs that find real regressions and marking them as high priority. ... Continue reading ->

[Next page](<https://devfeed.tech/topics/test-impact-analysis.md?cursor=WyIyMDE4LTA4LTI3VDE4OjE3OjE4KzAwOjAwIiwgIjg3YTc5ZDJhLTMyNGItNDZlNS04MTc0LTYzNjI2NTg2Yjg4YSJd>)