GitHub's gh stack Extension Makes Stacked Pull Requests Easy to Manage

·1 views

GitHub's gh stack extension, now available in public preview, allows developers to stack multiple branches so that individual pull requests contain smaller, more reviewable diffs. The tool automates the complex rebasing and force-pushing that previously made manually maintained branch stacks impractical for active development. Users need the gh CLI installed and authenticated, along with push access to their repository, to get started with a few simple commands. The extension includes a visual interface via gh stack view for navigating between layers and checking out branches without losing context. Because the feature is still in public preview, some rough edges remain, but it already handles branch propagation and PR linking automatically.

Read the full story at DEV Community

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

Discussion (0)

Docker Captain builds first SBX mixin kit to automate MLflow sandbox setup

Yhary, a Docker Captain and AI Engineer from Colombia, built his first Docker Sandbox Kit (SBX Kit) as part of a Docker Captains community activity. An SBX Kit is a declarative YAML file that automatically configures a sandbox environment at creation time, eliminating repetitive manual setup and configuration drift. He created a mixin kit designed to layer MLflow tracking server capabilities onto an existing Claude Code agent, enabling ML experiment logging and model versioning out of the box. The project followed a structured approach separating heavy dependencies into a Dockerfile from dynamic configurations kept in the YAML spec. The guide documents the full build process, including errors encountered, offering a practical reference for engineers looking to create reproducible AI development environments.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Five Reasons Teams Are Moving Away From Google Analytics in 2026

Google Analytics 4, which replaced Universal Analytics in 2023, introduced an event-based data model and a redesigned interface that many teams still find complex and time-consuming to navigate. Cookie-dependent tracking means GA4 requires consent banners in the EU and several other jurisdictions, potentially leaving up to half of actual traffic unaccounted for. The platform also applies data sampling on high-traffic queries, meaning reports can reflect estimates rather than precise figures. Privacy-focused regulations such as GDPR, CCPA, and Brazil's LGPD are tightening restrictions on cookie-based tracking, pushing teams toward cookieless analytics alternatives. Additionally, some teams are reconsidering whether sharing their site data with Google — an advertising company that benefits from that data — aligns with their priorities.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

SRE Cuts $150K+ in Annual Datadog Waste by Auditing Observability Configs

A senior site reliability engineer at a payroll-processing company discovered over $150,000 in annual waste hidden within their Datadog observability setup after conducting a thorough configuration audit. The largest single saving came from restricting the Sensitive Data Scanner to only log sources where personally identifiable information could realistically appear, cutting daily costs from $390 to $56 — an 87% reduction worth roughly $120,000 per year. Further investigation uncovered phantom services in the Service Catalog, created by a shared logging library that overrode service name tags, which skewed monitoring and incident response routing. The engineer found that observability spending tends to go unreviewed because teams treat it like insurance, allowing configuration drift and dead settings to accumulate unchecked. The key takeaway is that organizations should audit their observability configurations — not just their bills — regularly, asking whether each component still serves a real, current purpose.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Two-Stage Invite Flow Keeps Unverified Users Out of Auth Systems

A recommended pattern for B2B SaaS platforms holds invitations in a pending state and only creates a user record after identity verification is confirmed. The approach treats an invite as a time-limited capability rather than a user placeholder, preventing unverified email addresses from entering authorization queries. A four-state transaction model — issued, claimed, verified, and consumed — uses a hashed single-use token to prevent duplicate accounts and enumeration attacks. The final acceptance step should be executed as a single atomic database transaction, ensuring that race conditions or retries cannot produce duplicate memberships or orphaned records. For GDPR compliance, the same state machine should revoke all active sessions and API keys before an account is permanently deleted.

0 comments Read more at DEV Community