New codec misa77 claims 2x faster decompression than LZ4 with better ratios

·2 views

A developer has released misa77, an experimental data compression codec designed to outperform LZ4 in decompression speed. Benchmarks on the Silesia corpus show misa77 achieving up to 5,219 MB/s decode throughput, roughly twice the speed of LZ4's 2,505 MB/s. The codec also offers comparable or better compression ratios than LZ4, though its compression speed is significantly slower. The performance gains are attributed to reduced branching and a format optimized for out-of-order CPU execution on Intel x86-64 hardware. The project was shared on Hacker News by its creator, who spent several months developing it.

Read the full story at Hacker News

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

GitHub Launches Agentic Autofix for Code Scanning Alerts in Public Preview

GitHub released an agentic autofix feature for code scanning alerts in public preview on July 10, 2026, designed to automatically generate patches for identified vulnerabilities. The system moves each alert through multiple stages — from candidate patch generation through build, testing, security checks, and human review before any merge. Experts caution that simply measuring the percentage of alerts with a generated patch is a misleading success metric, as generation is only the first of many critical steps. A structured evaluation framework is recommended, using deterministic canary assignments, detailed failure classification, and predeclared reliability gates to ensure patches genuinely resolve security issues without introducing regressions. Broader deployment should be justified by bounded evidence and explicit safety checks rather than surface-level averages.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

GitHub Adds Security Review Feature to Copilot App, Raising Research Questions

GitHub announced on July 14, 2026, that security review functionality is now available within the GitHub Copilot app. A proposed research protocol has been outlined to evaluate how effectively users engage with this feature, focusing on whether reviewers inspect evidence, challenge findings, and appropriately escalate uncertainty rather than simply accepting suggestions. The protocol uses synthetic repositories with deliberately mixed findings — some relevant, some misleading, and some omitted — to test decision-making quality across different reviewer roles, including specialists, maintainers, and less-experienced users. Key metrics include how often unsupported claims are challenged, whether evidence is cited in decisions, and how correctly participants escalate unresolved issues. Researchers caution that acceptance rate alone is not a valid success measure, as low acceptance could reflect healthy skepticism, and that lab conditions cannot fully replicate real-world security review pressures.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

GitHub Mobile Can Now Trigger Copilot to Resolve PR Merge Conflicts

GitHub announced on July 8, 2026 that its Mobile app can initiate a Copilot cloud-agent workflow to address pull-request merge conflicts. The process follows a structured sequence: a mobile user delegates the task, the agent proposes a patch within defined boundaries, and a human performs the final merge after review. Experts caution against assuming a single tap fully resolves conflicts, stressing that the agent's scope should be explicitly limited to specific file paths while blocking changes to workflows, deployment configs, and policy files. Before handing off to the agent, users should confirm repository details, branch revisions, expected file scope, required checks, and the designated reviewer. Passing CI checks alone is insufficient — a human must semantically review each conflict resolution choice before the patch is merged.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

GitHub Expands Copilot BYOK Support in JetBrains With Accessibility Test Plan

GitHub announced on July 14, 2026, that Copilot for JetBrains has expanded its bring-your-own-key capabilities. Alongside this update, a proposed accessibility test plan was outlined to evaluate keyboard and screen-reader usability for the provider and model selection workflow. The plan treats provider and model selection as a dependent workflow rather than two independent dropdowns, meaning a chosen provider determines which models become available. Key testing criteria include stable focus during dynamic refreshes, clear announcements of state changes, and the prevention of invalid provider-model pairs from being submitted. The plan is not a review of existing defects but a structured matrix to guide future accessibility validation across various interaction, announcement, and state-correctness scenarios.

0 comments Read more at DEV Community