SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

Cash-on-delivery orders in Egypt turn delivery apps into accounting systems

Many online orders in Egypt are still paid in cash to couriers at the door, so money bypasses payment gateways and must be tracked manually. The author, founder and CTO of the logistics platform Rahwan run by Giantrex, describes the difficulty of tracking funds among vendors, couriers, and the platform. Each party is given a wallet, and an append-only transactions table records every movement with its reason and a balance snapshot. When an order is delivered, credits and debits across the three wallets must net to zero, which lets any discrepancy be traced to a specific order.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Terence Tao Launches Community Hub for Revising OpenAI's Math Write-Ups

Mathematician Terence Tao has announced a community hub called oai-math-revised, aimed at revising write-ups of mathematical work produced by OpenAI. The project invites community participation in reviewing and improving those documents. The announcement was posted on Tao's blog and shared on Hacker News, where it drew modest early attention. The source material provided is limited, so details such as specific revisions, timelines, and governance of the hub are not described here.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Quy trình bốn bước giúp tóm tắt tài liệu kỹ thuật dài bằng AI mà không bỏ sót chi tiết

Bài viết chia sẻ cách dùng chatbot AI để tóm tắt tài liệu kỹ thuật dài như RFC hay tài liệu API mà không làm mất các chi tiết quan trọng. Thay vì cắt văn bản theo số ký tự cố định, tác giả khuyên nên chia theo từng mục như xác thực, endpoint, mã lỗi và giới hạn. Mỗi phần được trích xuất theo một yêu cầu cụ thể, chẳng hạn liệt kê tham số bắt buộc, giới hạn hệ thống và những điều tài liệu không nói rõ, đồng thời yêu cầu AI chỉ dùng thông tin có trong văn bản. Các kết quả sau đó được gộp thành bảng tra cứu nhanh, đánh dấu những mâu thuẫn giữa các phần vì đó thường là nguồn gốc của lỗi. Cuối cùng, người đọc nên kiểm tra chéo vài dòng với tài liệu gốc và kiểm tra chính sách công ty trước khi đưa tài liệu nội bộ vào công cụ AI công cộng.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

TornadoVM 7 Lets Java Code Run on GPUs and FPGAs, GrapheneOS Gains Bank App Support

TornadoVM 7 has been released, dynamically compiling standard JVM bytecode into OpenCL, CUDA, or SPIR-V so Java programs can execute on GPUs and FPGAs. The update improves memory handling and reduces data transfer overhead between host and device. Separately, about a dozen major banking apps have started explicitly allowing verified GrapheneOS installations despite strict hardware attestation and Play Integrity checks. This lets privacy-focused Android users keep the hardened operating system without giving up mobile banking. The roundup also highlighted OBS Studio 33 and a community remaster of the 1990s game One Must Fall 2097.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Loan Match Uses One AI Call to Match 10,000 People With Personal Loans

A developer built Loan Match, a system that takes a CSV of people, reads lender websites daily, and emails each person a tailored list of loans they may qualify for, with reasons. The design began from fixed requirements rather than tool choice, and the file bypasses the API entirely: the browser uploads directly to S3 in checksummed chunks via presigned URLs, since API Gateway's size and timeout limits would fail on large files. A queue sits between S3 and the Lambda processor, providing visible retries, a dead-letter queue for failures, and a concurrency cap to protect the database. The Lambda streams rows in constant memory, normalizes messy spreadsheet data, and bulk-loads valid records into PostgreSQL, rejecting the whole import if more than 20% of rows are invalid. The author reports that the Go rewrite loads one million rows in about six seconds.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Supermicro Unveils 4U Server With 160 Hot-Swap NVMe Drive Bays

At FMS 2026, Supermicro showcased the ASG-4116S-NU160R, a 4U server that houses 160 NVMe SSD bays. The system uses standard 2.5-inch U.2 drives rather than a newer EDSFF form factor, which far exceeds the typical 24 to 48 drives found in comparable high-density servers. Its two-level slide-out tray separates the hot-swap drives from the CPU, memory, and networking, while central PCIe switches manage the connectivity that a single CPU could not provide directly. Each bay runs at PCIe 4.0 x1, and power and cooling are handled by redundant 2600W supplies and rear fans. The design aims to reduce server count, ports, and power overhead for large flash storage deployments.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Raspberry Pi 5 RAM Upgrades Blocked by Firmware Since 2024, Documentation Now Surfaces

Hobbyists had long upgraded Raspberry Pi 5 memory by swapping the LPDDR4X chip, a practice that ended when bootloader EEPROM releases dated 2024-09-23 and later began rejecting memory that differs from factory provisioning. Boards with swapped chips now fail SDRAM initialisation and will not boot, signalled by eight flashes of the green ACT LED. Raspberry Pi Ltd, the trading company, says the restriction targets resellers who fitted larger chips to low-RAM boards and sold them as higher-capacity models. The behaviour was only publicly documented in September 2026 by Jeff Geerling, while the official changelog does not mention a memory check, and a 2025 forum issue offers the clearest evidence.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Signal Adds Optional Phone-Number-Free Sign-Up, Paid via Zero-Knowledge Credentials

Signal has introduced an optional registration path that does not require a phone number, currently in its Android 8.28.0 beta released 16 September 2026. Accounts are identified by an Account ID and Account Key rather than any verified personal detail, and a one-time in-app purchase of about 3 USD in the US serves as the anti-spam measure. Zero-knowledge proofs, reused from Signal's donation badge system, are used so the payment cannot be linked to the new account. Google Play still records the purchase on the user's side, a limitation Signal acknowledges, and additional payment methods are planned. The feature is Android-only and unavailable on devices without Play Services for now.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

NCFA Outlines Five Commercial Paths for Developers Building on Open Banking

The National Crowdfunding & Fintech Association of Canada has identified five commercial opportunities for financial software developers in open banking. The UK processed 351 million open banking payments in 2025, which the article cites as evidence of measurable demand. The proposed opportunities include financial intelligence tools for lenders, consent and compliance infrastructure, embedded finance for business software, and programmable payments. Canada is implementing consumer-driven banking in stages, so read-access capabilities can be used now while write access and payment initiation come later. The NCFA says developers entering this space need to start with the customer and financial process rather than just the technical documentation.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Session Management II: Adding Context Compaction to a Basic AI Agent

This tutorial continues a series on building a basic AI agent from scratch, focusing on session management. Long-running sessions eventually exceed the model's context window and become slow and costly, so the author adds context compaction. The harness estimates token usage before each request and automatically compacts once it passes 80% of the token budget, or on demand via a manual command. Compaction splits the conversation at the latest user-turn boundary so tool calls and results stay paired, then summarizes older turns with a separate, low-temperature LLM call while the full transcript remains on disk. The summary is inserted as a clearly labeled untrusted-data user message to reduce prompt-injection risk, and the summarization call's usage is tracked in the session's cost estimate.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

14 Places to Launch Developer Tools in 2026, From Weekly Launchpads to Directories

A founder has compiled a shortlist of 14 platforms where developers can share new tools such as APIs, CLIs, SDKs, and open-source libraries in 2026. The list spans weekly launch competitions, technical communities like Show HN, and open-source discovery directories. Each option is described by its best use case and its trade-offs, since some suit an initial launch while others help projects be found long after release. The author discloses that their own platform, SemiLaunch, is the top pick, so the list is a practical founder's perspective rather than an independent popularity ranking. The piece also notes that listing terms and costs change, so creators should check each site's current submission rules.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer asks which part of shipping software takes longest beyond core features

A developer reflects on how much software development time goes into work outside the core product, such as authentication, databases, infrastructure, integrations, and deployment. They say they built Mudbase BaaS to address some of these backend challenges. They also mention exploring how AI-assisted development and code execution could simplify parts of the process. The post invites other builders to share which stage, from prototype to working application, they find most frustrating.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developers Build Linux GPU Driver for M4 Mac mini in a Month Using Coding Agents

Two developers, Cody Ho and Niklas Sheth, released a working OpenGL ES 3.0 conformant Linux GPU driver for the M4 Mac mini and MacBook Neo in September 2026. Most of the code was written by LLM coding agents, with Codex handling kernel and firmware reverse engineering and Claude handling user-space work. The driver ships in Gravity Linux, a Fedora remix forked from Asahi Linux, whose AI policy permits generative AI under disclosure rules while Asahi's forbids it. The team reports building the driver from live hardware traces rather than Apple binaries, reusing Asahi's kernel UAPI while writing new code, including a custom shader compiler. The project also relied on Khronos conformance tests to ground the agents, though it required notable human judgement, such as changing capture methods after an agent spent over a week on an unreplayable trace.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

How Booking.com Turned Marketing Into Computational Infrastructure

An article traces how Booking.com shifted marketing from persuasive creative work to a component of its computational systems. The company used Google AdWords to target user search intent precisely and pursued a long-tail strategy. By automatically generating thousands of landing pages and testing them at scale, it relied on controlled experiments rather than managerial intuition, a shift the piece describes as replacing the HiPPO approach. This linked search intent to specific accommodation offers and reduced market uncertainty. The result, according to the article, is a streamlined transactional customer journey where advantage comes from how efficiently decisions are structured and analyzed.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Google Accused of Illegally Felling 250,000 Trees in Finland for Data Center

A report says Google illegally cleared roughly 250,000 trees in Finland to make way for a data center. The clearing reportedly angered local residents, who saw the forest removed without proper authorization. The article, published by the New York Post in October 2026, is based on reporting about the project. The details provided in the source text are limited, so specifics such as the exact location, legal findings, and Google's response could not be confirmed here. The story gained attention on Hacker News, where it drew discussion about the environmental and community impact of data center construction.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Independent test finds 8 of 16 USB Coding Expert demo snippets run, the other 8 fail

A developer ran all 16 code snippets produced by the demo generator for USB Coding Expert, a WorkflowAlpha product. Eight snippets, covering the reverse, file and HTTP presets in Python, JavaScript and Bash, executed and returned the expected output. The remaining eight failed in three ways: the Python token preset returned Solidity syntax, the JavaScript token preset needed an uninstalled web3 dependency, and the Bash token preset required solc and geth, which were absent on a standard host. The four Solidity snippets could not be verified because no Solidity runtime was tested. The author notes the demo is a real, runnable code table, but says its marketing overstates what it does and omits the required toolchains.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer builds in-browser GIF upscaler, fixing halo artifacts and pixel art issues

A developer who runs a video tools site extended their WebGPU video upscaler into a browser-based animated GIF upscaler. They wrote a custom GIF decoder in TypeScript instead of relying on ImageDecoder, which lacks broad browser support and clamps short frame delays. Transparent pixels in GIFs store dark RGB values that caused dark halos when upscaled, so the developer fills those pixels with colors from opaque neighbors before processing. Alpha is handled separately with bilinear sampling and a 50% threshold to keep edges smooth. For pixel art, a nearest-neighbor mode at whole-number scales from 2x to 8x preserves hard edges and exact colors.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Three Years of Self-Hosting: What Worked and What Wasted Time

A developer shares a candid three-year account of self-hosting, drawing on personal experience rather than offering a tutorial. They favor free-tier services like Cloudflare Pages, Workers, and D1 for small projects, noting a subrequest limit that tripped up an early health checker. The author found a larger VPS unnecessary and suggests upgrading only once utilization stays above 70 percent. Kubernetes was judged too costly to learn for a small homelab, with Docker Compose, backups, and Uptime Kuma delivering most of the benefit. The piece also covers outsourcing email deliverability and buying domains only after a live page exists.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Node Core OS byLAEV Proposes Verifiable, Person-First Digital Identity

Developer Lerry Alexander Elizondo Villalobos outlines Node Core OS byLAEV, a project whose foundation is personal human identity. He argues that cryptographic keys, identity documents, and institutional statements should not be equated with the person or treated as infallible. The proposed method starts from first principles and turns ideas into verifiable rules and reproducible tests. A peer-to-peer network would let nodes cross-check evidence rather than defer to institutional authority or majority opinion. An initial stage would focus on demonstrating the identity core's functionality and transparency, with zero-knowledge proofs considered later for privacy-preserving verification.

0 comments Read more at DEV Community
← NewerPage 32 of 2879Older →