SShortSingh.

Programming

0
ProgrammingDEV Community ·

Firm Found No Wipe Certificates After 340 Leased Laptops Were Returned

A logistics and asset management firm discovered a critical data security gap after 340 leased laptops were returned to their lessor last spring without verified sanitisation records. An auditor's question revealed that while the company maintained a careful chain of custody for the physical devices, no process existed to confirm that data stored on them had been wiped. The problem extended further: 28 laptops were never returned by users, pre-2021 devices lacked full-disk encryption, and four multifunction printers were sent back on a separate lease with drives that had been caching scanned finance documents. In response, the company overhauled its disposal process so that sanitisation now occurs before collection, with destruction certificates reconciled against the asset register on a per-serial-number basis. Lost devices are now treated as security incidents rather than simple financial write-offs, and all storage-bearing equipment — including printers and network devices — falls under the revised policy.

0
ProgrammingDEV Community ·

CI Secret Exposed for 14 Months After Base64 Encoding Bypassed Log Masking

A software team discovered their container registry deploy token had been visible in CI build logs for fourteen months after an engineer spotted it during a failed build review. The leak occurred because the CI runner's masking system only redacts exact string matches — once the token was base64-encoded for a basic auth header and the shell ran in trace mode, the transformed value was logged unmasked. A second exposure came from the registry returning full request headers in an error response, which the team logged wholesale. To remediate, the team rotated the token, registered all derived secret values explicitly with the runner, banned shell tracing in credential-handling steps, and replaced the long-lived token with a short-lived OIDC credential expiring in 15 minutes. A follow-up automated log scan uncovered a second leaked secret in a separate repository tied to an unowned service.

0
ProgrammingDEV Community ·

How to Spot Early Warning Signs of Company Trouble Before Layoffs Hit

Workplace advisor Asael Shinder argues that companies typically show subtle distress signals for two to three quarters before any official bad news is announced. Common indicators include unfilled vacancies, contractor cuts, shrinking roadmap horizons, unexpected finance requests, and tightened budgets for travel and training. Shinder cautions that any single sign means little, but four or more appearing within the same quarter likely indicate a pattern worth taking seriously. Rather than panicking or disengaging, he recommends quiet, low-cost steps such as updating your CV, alerting your professional network, and reviewing your contract terms. He emphasises that being prepared carries no downside — an updated CV and warm network are valuable regardless of whether the company stabilises or not.

0
ProgrammingDEV Community ·

Why Streaming an Upload Requires an End-to-End Architecture, Not Just a Stream API

Accepting a stream in an upload handler does not guarantee the entire file is never held in memory — it depends on how bytes travel through every layer of the pipeline. A common pattern involving browser files routed through an interactive UI circuit, converted to byte arrays, encoded as Base64, and wrapped in JSON can cause multiple full-file copies to overlap in memory simultaneously. Base64 encoding alone inflates binary data by roughly one third before accounting for the surrounding JSON and transport overhead, making this approach a risk for large or concurrent uploads. A more robust architecture routes the browser's multipart HTTP upload directly to a dedicated endpoint, which passes a stream and declared length straight to storage without intermediate conversions. Even then, frameworks like ASP.NET Core may buffer smaller files in memory and larger ones on disk, so claims of zero-copy or constant-memory usage are rarely fully accurate.

0
ProgrammingDEV Community ·

Software Apprentice Builds 16-Agent Distributed Coding Setup Using Old Mini-PCs

Moses Amani, a software apprentice at Zero One, has shared how he built a multi-machine agentic workflow setup after being thrust into agent-driven development from the start of his role. To avoid crashing a single machine with 16-plus simultaneous AI agents, he repurposed old mini-PCs as headless Unix nodes, linking them into a private mesh network using Tailscale. A tool called Herder handles orchestration across all agent streams, alerting him via dashboard and phone whenever an agent completes a task or requires human input. Amani uses LazyGit, Hunk, and Neovim to review code changes in context before they are integrated. While the setup dramatically increases code output, he cautions that raw line counts are a misleading metric and that high throughput demands equally rigorous human validation.

0
ProgrammingDEV Community ·

AI Red-Teaming and Observability Are Merging Into One Continuous Practice

Engineering teams building large language model features have traditionally treated pre-launch red-teaming and post-launch observability as separate disciplines, but that divide is increasingly collapsing. Real users probe LLM systems far more creatively and persistently than any scheduled red-team exercise can anticipate, meaning vulnerabilities often surface only after a product ships. Unlike conventional software, LLM failures are semantic rather than structural — a jailbroken model returns a successful response with harmful content rather than throwing an error, making standard monitoring tools insufficient. Teams are responding by adopting continuous automated adversarial testing, logging intermediate reasoning steps in agent chains, and deploying lightweight classifier models to flag policy violations in live outputs. The practical implication is that security testing is no longer a pre-launch phase to be completed, but an ongoing operational process that overlaps directly with production monitoring.

0
ProgrammingDEV Community ·

Why Collecting Less Data Now Protects Your Team Later

A data privacy perspective shared on DEV Community argues that organizations routinely collect far more data than they actually need, creating hidden legal and security risks. The author warns that stored data becomes a liability during breaches, as exposed records carry their current sensitivity regardless of when they were collected. Subject access requests and due diligence processes compound the problem by requiring teams to account for every copy of data held. The piece recommends assigning retention periods to every database table, automating scheduled deletions, and logging only the fields genuinely required. As data regulations tighten globally, the author concludes that teams practicing minimal data collection today will be better positioned for compliance tomorrow.

0
ProgrammingHacker News ·

Why Papua New Guinea Captivates One Writer's Imagination

A writer published a personal essay on Substack exploring their ongoing fascination with Papua New Guinea. The piece, shared on Hacker News, attracted points and reader discussion. The article reflects on what makes the country so intellectually and culturally compelling to the author. Papua New Guinea is known for its extraordinary linguistic diversity and largely unexplored landscapes, which often draw outsized curiosity. The post resonated enough with readers to be shared and upvoted on the tech-focused news aggregator.

0
ProgrammingDEV Community ·

Developer Fine-Tunes Open LLMs to Support Moroccan Arabic Darija and Arabizi

A developer has released two open-source language models, SILMA-9B-Darija and SILMA-2B-Darija, specifically fine-tuned to handle Moroccan Arabic (Darija) and Arabizi, the Latin-script form of Arabic that uses numbers for certain sounds. The models were trained on a dataset of over 46,000 cleaned Darija instruction and conversation pairs drawn from real Moroccan exchanges and cultural content. Benchmarks run on a 100-question test covering local history, proverbs, and code-switching showed both models outperforming existing Moroccan-dialect alternatives on accuracy and Darija fluency. Unlike standard Arabic models that default to Gulf Arabic or Modern Standard Arabic, the new models retain native Darija vocabulary and handle prompts in both Arabic script and Latin letters. Both models are publicly available on Hugging Face, with the 9B version requiring around 6 GB VRAM and the 2B version running on under 3 GB, making it suitable for consumer devices.

0
ProgrammingDEV Community ·

Why coding knowledge still matters in the age of AI-generated code

A software developer argues that while AI tools can generate code faster and often cleaner than humans, the critical skills of evaluating correctness, spotting hidden flaws, and understanding trade-offs remain entirely human responsibilities. The author contends that cheap code generation has not eliminated complexity but shifted it — making careful code review more costly and easier to skip, not less important. Developers who cannot read code risk silently accepting poor structural decisions, swallowed errors, or mismatched design patterns embedded in AI output. The piece highlights specific danger zones including debugging, security vulnerabilities, and architecture, where AI-generated solutions can appear locally correct while being globally flawed. The core argument is that coding literacy is no longer about typing speed but about maintaining the judgment needed to verify, challenge, and own what gets built.

0
ProgrammingDEV Community ·

12 critical questions to ask vendors before buying a text-to-SQL tool

A technical guide published on DEV Community outlines 12 evaluation questions buyers should ask text-to-SQL vendors before committing to a purchase. The article argues that product demos are misleading because they typically showcase clean, small schemas, while real-world databases are far more complex and error-prone. Key concerns raised include how vendors measure accuracy, what verification badges actually certify, and whether identical questions phrased differently return consistent results. The piece warns that wrong answers from such tools are especially dangerous because end users, particularly in finance, often cannot distinguish correct outputs from plausible-looking incorrect ones. Buyers are advised to test vendors live during meetings using questions with known answers rather than relying on polished demonstrations.

0
ProgrammingDEV Community ·

30-Day AI Self-Review Experiment Missed a Bug a Human Caught in 5 Minutes

A developer ran a 30-day experiment in which an AI model reviewed every pull request it generated, supported by static analysis and linting tools. The setup performed reliably for 29 days, catching minor issues and enforcing coding standards. On day 30, a human engineer identified a race condition in under five minutes — a logic flaw tied to real-world user behavior that the AI had repeatedly overlooked. The bug stemmed from a mismatch between the system's assumptions and actual user actions, something the AI could not detect because it lacked domain context and understanding of intent. The author concluded that AI review works best as a complement to human review, handling repetitive checks while humans focus on logic, design decisions, and user impact.

0
ProgrammingDEV Community ·

Developer Builds Browser-Based Dataflow Engine Where the Graph Executes the Program

A developer has created an experimental browser-based tool called Computational World (CW), also known as Simwire, which treats a directed graph as both the visual interface and the actual execution model. Unlike traditional visual programming environments that hide computation inside high-level nodes, CW exposes the underlying structure so users can inspect and manipulate it directly. The system resolves execution order automatically using Kahn's topological sorting algorithm, ensuring nodes only run after their dependencies have produced outputs. CW is designed around primitive operations that can be composed into more complex behaviors, rather than relying on pre-built black-box components. A key goal is live propagation — changing a parameter on the canvas immediately updates all dependent nodes, eliminating the repeated write-run-inspect cycle of conventional programming.

0
ProgrammingDEV Community ·

Fine-Tuning, RAG or Prompting: Picking the Wrong AI Method Can Cost Thousands

Enterprises frequently default to fine-tuning AI models when simpler, cheaper alternatives would better solve their problem, according to a cost-focused analysis. The three main approaches — prompting, retrieval-augmented generation (RAG), and fine-tuning — each address a distinct category of issue rather than being interchangeable options. Prompting is best for shaping tone and format, costs little beyond token usage, and can be adjusted instantly in production. RAG is the appropriate choice when a model lacks access to specific or up-to-date information, typically costing £15,000–£60,000 to build. Fine-tuning, the most expensive option at £20,000–£80,000, is suited to teaching consistent behavioural patterns but is ineffective at injecting factual knowledge, which it cannot update, audit, or reliably cite.

0
ProgrammingDEV Community ·

How Kubernetes Schedules Pods: Filtering, Scoring, and Manual Placement Explained

Kubernetes scheduling is the process of binding a pod to a node by setting the spec.nodeName field, after which the kubelet on that node runs the pod. The kube-scheduler automates this in two steps: first filtering out nodes that lack sufficient resources or have incompatible taints, then scoring the remaining eligible nodes to pick the best fit. Developers can bypass this process entirely by manually specifying a node name, but doing so skips critical safety checks around resources, node existence, and taint compatibility. Once a pod is scheduled, it remains on that node for its lifetime, with the sole exception being a NoExecute taint, which can evict a running pod that lacks the required toleration. Manual scheduling is useful for demos and edge cases but is not recommended as a standard practice in production environments.

0
ProgrammingDEV Community ·

Security Analysis Flags Nine High-Risk Vulnerabilities in Grove Finance's $1.9B DeFi Protocol

A senior DeFi security researcher published a vulnerability surface analysis of Grove Finance, a multi-chain yield-aggregation protocol managing approximately $1.92 billion in total value locked across Ethereum and several Layer 2 networks. The assessment, dated September 15, 2026, identified nine distinct high-impact attack vectors and assigned the protocol an overall risk score of 7.4 out of 10. Among the most critical findings were a re-entrancy flaw in the vault deposit and withdrawal functions, an untimelocked proxy upgrade mechanism vulnerable to multisig compromise, and a governance system with only a six-hour proposal delay. Additional risks include oracle feed manipulation, flash-loan exploitation of the rebalancing loop, and weaknesses in the protocol's cross-chain bridge components. Researchers recommended immediate remediation of the highest-severity issues alongside a structured roadmap to address the remaining vulnerabilities.

0
ProgrammingDEV Community ·

LifeOps Part 3: Sport, Library, Cinema and Kanban Modules Added to Personal OS App

Developer project LifeOps has reached its third milestone, introducing four interactive daily-use modules to the personal operating system web application. The new modules cover physical fitness tracking with dynamic metrics, intellectual leisure through book reading progress and an audiovisual catalogue, and professional productivity via a four-column Kanban board linked to projects. A dual-view toggle — switching between an inspirational card grid and a condensed data table — was implemented across all modules, with the user's preference saved persistently in localStorage. The Kanban board stores a chronological activity log using PostgreSQL's JSONB format, while the sports module supports multiple disciplines including running, cycling, gym, and swimming. The live application is accessible at datalaria.com, with two further instalments planned covering report generation and zero-cost cloud deployment.

0
ProgrammingDEV Community ·

Developer Builds Independent Kernel-Level Tests to Verify AI Agent Code Claims

A developer, inspired by researcher Hemapriya Kanagala's question about AI benchmark reliability, built a small native backup client with eight distinct behaviors to independently verify whether an AI agent's code actually does what it claims. Rather than trusting the agent's own logs or test results, the experiment observes program behavior from outside — using packet captures, kernel-level activity, and file comparisons. The setup uses a C client, a Python receiver over loopback TCP, and SHA-256 digest checks to confirm that saved files genuinely match their source. The author notes that when an agent produces both code and its tests, shared misunderstandings between the two can make passing tests misleading. The experiment argues that execution traces, network activity, and resulting file state together provide independent evidence to judge a program's behavior against its actual requirements.

0
ProgrammingDEV Community ·

HTTPS Exposes Website Hostnames in Plain Text Due to TLS Design

Every HTTPS connection begins with a TLS ClientHello message that includes the destination hostname in plain, unencrypted text via an extension called Server Name Indication (SNI). A developer confirmed this by capturing raw TLS handshake data and extracting readable hostnames — such as 'login.mybank.example.com' — without decrypting a single byte. The exposure is not a flaw but an intentional protocol design: servers sharing one IP address need to know which site is being requested before they can present the correct certificate and establish encryption. This creates a chicken-and-egg problem where the hostname must be sent before any encrypted channel exists to protect it. A proposed standard called Encrypted Client Hello (ECH) aims to address this, but it has not yet seen widespread adoption.

0
ProgrammingDEV Community ·

Dev Builds URL Shortener to Learn Backend Basics: Databases, Redis, and HTTP

A developer learning backend fundamentals built a URL shortener as a starter project, finding it an effective way to explore core concepts in a manageable scope. The project involved generating unique short codes using nanoid, storing URL pairs in a database, and handling HTTP 302 redirects to route users from short links to original URLs. To prevent abuse, Redis was integrated as a rate limiter, using fast in-memory counters tied to each alias with automatic expiry windows. Collision handling was addressed through a retry loop that regenerates a new code if a duplicate alias is found in the database. The full source code has been shared publicly on GitHub as a reference for other developers tackling similar beginner backend projects.

← NewerPage 1301 of 3289Older →