SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

Why Your A/B Test Results May Be Wrong Before You Even Read Them

A technical guide published on DEV Community outlines three critical validity checks that should precede any interpretation of A/B test results. The first is an A/A test, where two identical experiences are run against each other to expose tracking bugs or randomization errors in the testing pipeline itself — Microsoft's experimentation team once found a false-positive rate of 30% this way. The second check is Sample Ratio Mismatch (SRM), which uses a chi-squared test to verify that traffic was actually split as intended, since even a 4% imbalance can signal bot skew, ID collisions, or broken redirects. The third issue flagged is Simpson's Paradox, where aggregate results can contradict segment-level findings, making the overall metric misleading without deeper breakdown. The article argues these infrastructure checks are not one-time pre-launch steps but should run continuously so pipeline regressions are caught before they corrupt live experiment data.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Django 6.1's FETCH_PEERS cuts N+1 query loops from 2,001 queries to just 2

Django 6.1, released on 5 August 2026, introduced a queryset-level setting called fetch_mode with a FETCH_PEERS option designed to automatically batch foreign key lookups without requiring explicit select_related() or prefetch_related() calls. In benchmark tests using 2,000 book records across 50 authors on a local Postgres 17 instance, the default N+1 approach fired 2,001 queries and took 1,144.6 ms, while FETCH_PEERS reduced that to just 2 queries and 13.1 ms — nearly matching select_related performance. The feature works by issuing a single WHERE id IN (...) batch query on first access, functioning like an on-demand prefetch_related(), and also handles deferred fields beyond just foreign keys. However, testing revealed a documented limitation: FETCH_PEERS does not apply to reverse foreign key managers, meaning a loop over author.books.all() still produces the same number of queries regardless of the fetch_mode setting on the parent queryset.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Unity Tool Lets Developers Scan All Prefabs for Missing Scripts Before Build

Missing scripts in Unity prefabs often go undetected during development, surfacing only after a script is renamed, deleted, or affected by a merge conflict. Unity flags these broken references in the Inspector as 'Missing (Mono Script)', indicating a serialized component whose backing script can no longer be resolved. A developer on DEV Community has shared an Editor utility called MissingScriptScanner.cs that automates detection by scanning every prefab asset in a project using PrefabUtility.LoadPrefabContents. The tool reports affected prefab paths and a total count of missing scripts directly in the Unity Console via a Tools menu command. While a missing script does not always cause an immediate crash, it can silently strip critical logic such as collision, AI, or animation behaviour from a prefab.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Langfuse fills the observability gap left by traditional APM tools for AI agents

Conventional observability platforms like Grafana and Datadog excel at monitoring infrastructure and HTTP performance but cannot detect AI-specific failures such as hallucinations, wrong tool selection, or poor response quality. These semantic and reasoning-level issues leave no trace in standard metrics, making it nearly impossible to diagnose why an AI agent underperformed in a specific interaction. Langfuse is an open-source LLM engineering platform that addresses this blind spot through tracing, prompt management, evaluation, and experimentation features. Unlike general APM tools, it treats AI applications as continuously evolving systems that improve through iterative feedback loops driven by user signals and prompt version comparisons. The platform also solves the problem of hardcoded prompts by centralizing them with native versioning and environment-based deployment controls, removing the need for full code deployment cycles on every prompt adjustment.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Developer Builds 386 PC Emulator Running on RP2350 Microcontroller

A developer has created 'frank-386', an open-source project that emulates an Intel 386 PC on the RP2350 microcontroller. The project is hosted on GitHub under the handle rh1tech and allows the low-cost RP2350 chip to run software designed for classic 386-era computers. The RP2350 is a microcontroller from Raspberry Pi, known for its affordability and versatility in embedded projects. This kind of emulation project demonstrates the growing capability of modern microcontrollers to replicate legacy computing hardware.

0 comments Read more at Hacker News
0
ProgrammingHacker News ·

EuroBirdPortal Tracks Live Bird Movements Across Europe in Real Time

EuroBirdPortal is an online platform that visualizes bird movements across Europe in real time. The tool aggregates bird observation data to display migration and distribution patterns continent-wide. It is accessible via the official EuroBirdPortal website and is aimed at researchers, birdwatchers, and conservationists. The platform provides a dynamic way to monitor avian activity across different seasons and regions. It was shared on Hacker News, drawing early attention from the tech and science community.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

How to Build a RAG System on AWS Using Terraform, Bedrock, S3, and OpenSearch

A technical guide published at the AWS Builder Center demonstrates how to implement a Retrieval-Augmented Generation (RAG) system entirely on AWS using Terraform as the infrastructure-as-code tool. The setup uses Amazon S3 as the document source, Amazon Bedrock Knowledge Bases to manage the ingestion pipeline, and Amazon OpenSearch Serverless as the vector database for storing and searching embeddings. During ingestion, Bedrock splits uploaded documents into chunks, converts them into vector embeddings using Amazon Titan Text Embeddings v2, and stores the results in OpenSearch Serverless with KNN similarity search powered by the HNSW algorithm and FAISS engine. At query time, a user's question is converted into an embedding, matched against stored vectors via cosine similarity search, and the retrieved context is passed to a large language model to generate a grounded answer with source citations. A Streamlit-based UI ties the workflow together, allowing users to upload documents, trigger knowledge base syncs, and submit natural-language queries interactively.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Local-First Multi-Agent Platform Using Strands, Ollama, and MCP Gateway

A developer has built a local-first agent platform in Python that allows a single AI agent to access travel, finance, and entertainment capabilities without being directly coupled to each tool's implementation. The architecture uses Strands Agents for orchestration, Ollama for running a local language model, and the Model Context Protocol (MCP) to define a clean tool contract between the agent and domain servers. A central MCP gateway sits between the agent and multiple focused domain servers, hiding backend topology and making it easier to add or update individual domains independently. The design enforces three key constraints: the model runs locally, the agent connects to one stable tool endpoint, and each domain owns its own small MCP server. A FastAPI backend and Streamlit interface handle user interaction, while agent profiles are managed through a central registry that separates agent behavior from tool implementation.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

GitHub Pages Sites Stay Public Even When the Source Repository Is Private

A developer exploring GitHub Pages discovered that making a repository private does not automatically make the published website private — the two settings are independent. On free and Pro personal plans, any GitHub Pages site remains publicly accessible to anyone with the URL, regardless of the repository's visibility. Access-controlled Pages sites are only available under GitHub Enterprise Cloud, and only for organization-owned repositories, not personal accounts. This means individual developers cannot restrict who views their Pages site simply by upgrading their GitHub plan. Those using Pages for internal documentation, staging environments, or unreleased client projects should be aware they need a separate access-control solution beyond GitHub's standard offerings.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Reflects on Identity Crisis After Shifting from Building to Judging AI Output

A software developer writing a six-part series on building with AI agents describes a growing psychological toll from spending entire workdays reviewing and approving AI-generated work rather than creating code independently. The developer recounts logging 40 weekly decisions — all marked 'confirmed' — without recalling the reasoning behind a single one, a sign of deepening disengagement. This shift from maker to quality-checker left the developer feeling hollow, a state distinct from ordinary tiredness and not relieved by rest. A brief moment of genuine engagement returned only when an outside commenter flagged a flaw, prompting hands-on debugging that restored a sense of purpose. The piece frames this experience as 'judging fatigue,' arguing that prolonged evaluation work, even when useful, erodes a developer's sense of authorship and creative identity.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

GPS location data survives image resizing in sips but not Pillow, audit finds

A web developer testing image-processing tools discovered that GPS metadata behaves inconsistently depending on the tool used to resize or convert photos. Apple's built-in sips utility retained GPS coordinates in all output formats it produced, including JPEG, PNG, and resized files, while Pillow stripped location data by default unless original EXIF was explicitly passed back. A particularly notable finding was that sips-generated PNG files stored GPS data twice — once in the EXIF chunk and once in an embedded XMP packet — meaning a scanner checking only EXIF would falsely report the file as clean. The developer built a directory scanner that checks both EXIF and XMP for GPS entries, returning an error if either contains location data or if a file cannot be read. Four remediation approaches were evaluated, with targeted GPS IFD deletion identified as the cheapest option for JPEG files since it avoids a full re-encode and preserves pixel data.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Cleora v2 WordPress Theme Rebuilt From Scratch With No Build Pipeline

HeliconiaIO has released Cleora v2, a complete ground-up rebuild of its minimal editorial WordPress theme focused on speed, accessibility, and simplicity. The original version relied on Tailwind CSS, Alpine.js, PostCSS, and an npm build pipeline, which the team concluded added unnecessary complexity for a small editorial theme. Version 2 replaces all of that with a single hand-authored stylesheet and roughly 5 KB of vanilla JavaScript, eliminating the need for any frontend build step. Design decisions are now managed through CSS custom properties and theme.json to keep the block editor and front end visually consistent. Cleora v2 is freely available on the official WordPress theme directory and its source code has been published on GitHub.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Free Browser Toolkit for File Viewing, Converting, and P2P Sharing

A developer has released WebTools Utility, a free and open-source browser-based application hosted on GitHub Pages. The tool combines a file viewer, file converter, ZIP support, and peer-to-peer file sharing in a single browser tab. The entire app consists of just three files — index.html, style.css, and worker.js — with no frameworks or build steps required. P2P transfers are powered by WebRTC via PeerJS, using 256KB chunks and four parallel senders for efficiency. The project is publicly available on GitHub and costs nothing to run, as it is fully hosted on GitHub Pages.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Docker PHP builds failing on servers? A seccomp and xz sandbox conflict explained

Developers running Docker-based PHP builds on certain Linux servers may encounter a reproducible failure caused by a conflict between recent versions of xz, older distribution kernels, and Docker's default seccomp profile. Modern xz attempts to activate a Landlock sandbox before processing data, but when the server kernel lacks support, Docker's seccomp filter returns EACCES instead of ENOSYS, causing xz to abort rather than proceed. The fix involves creating a custom seccomp profile that explicitly blocks Landlock-related syscalls and restarting the Docker daemon with that profile applied. Additionally, Docker BuildKit must be disabled using DOCKER_BUILDKIT=0, since the modern build engine bypasses the custom seccomp profile entirely. The author notes this is a temporary workaround rather than a true fix, and the custom profile should be removed once the server kernel is upgraded to one that supports Landlock natively.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer loses 48 hours to cross-device tempfile bug that broke atomic JSON writes

A software developer spent 48 hours debugging a silent failure in an atomic file-write helper used to safely update JSON config files. The bug stemmed from Python's os.replace, which is only atomic when the source and destination files reside on the same filesystem device. On a local machine, the default tempfile directory and the target path shared the same device, masking the issue entirely during testing. When deployed to a remote workspace with a different filesystem layout, the cross-device rename caused the operation to fail. The fix was straightforward: passing dir=path.parent to tempfile.mkstemp ensures the temporary file is created on the same device as the destination, keeping the rename truly atomic.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Oracle AI Database 26ai Introduces Deep Data Security for Row and Column Access Control

Oracle AI Database 26ai has introduced a feature called Oracle Deep Data Security (Deep Sec), a database-enforced authorization framework that controls access at the row, column, and cell level. Unlike traditional application-layer security, Deep Sec enforces access policies directly within the database, regardless of whether the request originates from a human user, a reporting tool, or an AI agent. The system introduces three new building blocks: local end users (lightweight identities that own no schema), data roles (fine-grained privilege carriers that can sync with external identity providers like Microsoft Entra ID), and data grants (SQL-native policy objects defining read/write permissions). Local end users are created with a new CREATE END USER statement and tracked via a dedicated DBA_END_USERS dictionary view, but they cannot connect independently and must be authorized through data roles and grants. The feature is designed to complement existing Oracle tools like Label Security and Data Masking while offering a cleaner, more scalable approach to least-privilege access enforcement.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How to Responsibly Integrate AI Into US Building Permit Workflows

A structured pipeline for using AI in US building permit processes treats AI outputs as unverified drafts that require human classification, documentation, and licensed professional review before submission. The approach involves logging metadata for every AI-generated output, including the model used, prompt, data sources, and a named review owner, to maintain clear provenance. AI is considered appropriate for early-stage tasks such as massing studies, facade options, and site summaries, while code-critical elements like egress, structural loads, and energy compliance remain under licensed human oversight. Before a licensed architect or engineer signs off, any unverified AI-generated geometry must be removed or replaced with checked, authoritative information so building officials can distinguish controlled documents from exploratory references. The framework acknowledges that ongoing federal AI governance guidance and unresolved US Copyright Office questions around AI outputs make thorough documentation more reliable than asserting ownership over generated content.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How Hardware Timestamping and PTP Eliminate Jitter in Distributed Systems

Software-based timestamps in distributed systems introduce variable delays ranging from microseconds to milliseconds due to interrupt handling, scheduler preemption, and NIC queues, making them unreliable for precision-critical applications. Hardware timestamping, by contrast, captures the exact moment a packet hits the network interface card, pushing timing uncertainty down to the nanosecond range. Systems such as high-frequency trading platforms, distributed databases, and PTP-synchronized networks depend on this accuracy to maintain correct event ordering and regulatory auditability. The IEEE 1588 Precision Time Protocol is specifically designed to leverage hardware timestamps and a hierarchical clock model to compensate for network-path delays. Tools like linuxptp, along with kernel interfaces such as ethtool and SIOCSHWTSTAMP, allow engineers to enable and validate hardware timestamping on commodity Linux systems.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer builds CLI tool to sequence Spotify playlists, learns hard lessons about API stability

A developer created 'setlisted', an open-source command-line tool that sequences Spotify playlists into structured sets with a warm-up, build, peak, and cooldown arc using only track metadata. The tool also generates discovery mixes without repetition and builds hour-targeted event sets to an exact duration. During development, the project hit a silent API breakage when Spotify renamed a field mid-response, leaving the old field populated with a boolean instead of its original object, causing logic errors that bypassed standard null and existence checks. This led the developer to adopt a principle of validating field types, not just presence, and building core functionality on durable metadata rather than rich fields that are more likely to be restricted. The project doubles as a case study in defensive API integration, with the author recommending diffing raw API responses when output goes subtly wrong without any errors thrown.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

DEV Community Guide Shows How to Fine-Tune LLMs Using Unsloth Studio

A new tutorial series on DEV Community walks developers through fine-tuning large language models using Unsloth Studio, covering the full workflow from data preparation to model training. The guide distinguishes fine-tuning from RAG, explaining that fine-tuning embeds new knowledge directly into a model's weights rather than supplying it as context at query time. The author demonstrates the process using a manually created PDF of 2026 FIFA World Cup facts, which was cleaned and converted into a ChatML-format conversational dataset. Dataset generation was handled either via Claude or ChatGPT prompts, or through Unsloth Studio's built-in Data Recipes feature. For training, the author selected the Llama-3.1-8B-Instruct model, citing its 8.6GB GPU memory requirement as a practical fit for Apple Silicon hardware with 32GB unified memory.

0 comments Read more at DEV Community
← NewerPage 1269 of 3231Older →