At 93, Dharam Singh Mann recalls his battles at Tokyo and a fierce final against Pakistan

Dharam Singh Mann, a gold medal winner at the 1964 Olympics, is now 93 years old and still vividly remembers his time in Tokyo. He has been recounting a fierce final match against Pakistan from those Games. The article presents his reflections on that intense contest, which remains a lasting memory from his career. The source provides limited detail beyond the core account of his recollection and the gold medal achievement. The piece focuses on how these past battles continue to live on in his memory decades later.

CVPR 2027 Submission Guide: Key Deadlines and New Rules

The CVPR 2027 conference has set three firm submission deadlines, all in Anywhere on Earth time, beginning with paper enrollment on November 10, 2026. Full paper submission is due November 16, 2026, with supplementary materials following on November 23, and organizers have stated no extensions will be granted. Two new rules apply this year: a limit on how many papers an author can co-author without an eligible reciprocal reviewer, and a single-paper cap for groups without prior accepted work at eligible venues, where ACL, EMNLP, and NAACL do not qualify. Submissions must also include an LLM Usage Statement disclosing any generative AI use, and omitting it can lead to desk rejection. Authors are capped at 25 papers each, and every author needs a complete OpenReview profile and enrollment to avoid rejection.

Slowing Audio to 0.8× Stretches an 8-Second Clip to 10 Seconds

A common mistake when changing audio speed is assuming a 0.8× setting shortens or lengthens the clip by the same percentage. Duration scales inversely with the speed multiplier, so an eight-second track slowed to 0.8× lasts 10 seconds, a 25% increase. The Creatune team tested this on November 4, 2026, using a synthetic WAV tone and a public speed changer. Their measurements covered exported file duration and format metadata, not listening quality or pitch preservation. The article gives the formula for percentage change and shows how to request a specific multiplier to hit a target length. It also recommends measuring the exported file rather than relying on the interface's displayed total.

New tool finds code by meaning, not keywords, using a free on-device index

A guide describes how grep fails when a developer doesn't know the exact term used in code, such as searching for a lockout feature named a cooldown or throttle. The osFoundry desktop app addresses this by converting code chunks into vectors using the free, open-source bge-m3 model running locally, so nothing is uploaded or charged. Its scan step previews what gets indexed and skips irrelevant folders, lockfiles, and large or binary files. Users can then ask questions in plain English, and the agent combines keyword and semantic search to return the relevant file and line range. The indexing is incremental, so editing a few functions only re-embeds those chunks.

Unshared Claude Code rules cause four team failure modes, researchers argue

A developer's use of Claude Code without team-wide rules leads to inconsistent output, since each person's habits and context shape the results differently. Fast code generation can hide requirement misunderstandings, especially when specifications pass through translation or intermediaries. Large AI-generated diffs that touch unrelated files overwhelm reviewers, who either skim and approve or reject them wholesale. The author recommends separating repository facts in CLAUDE.md from on-demand procedures in skills, since the full CLAUDE.md loads every session while skill bodies load only when matched. Rules also decay without versioning and a clear owner, so shared documents that nobody maintains lose their authority.

A Developer's Guide to Testing Beyond Happy-Path Checks

The article argues that software testing should go beyond confirming correct return values, since real failures like unavailable backends or interrupted operations also matter. It describes unit tests as isolated checks of functions, classes, or methods, covering valid and invalid inputs, boundary values, and dependency failures. Mocking and stubbing are presented as ways to isolate dependencies such as databases. Component tests are described as covering larger parts of an application, such as a UI screen or API service, including loading states, error messages, and stale responses. Integration tests then check whether connected components, such as a service and database or a UI and backend, work together correctly.

Edtech Password Reset Controls: Token Expiry, Session Revocation and Audit Logs

The piece argues that an education platform should only leave a managed identity provider if its password reset flow can invalidate a stolen session, enforce token expiry, and keep an audit trail that never stores the reset email link. It stresses that a successful password change alone does not prove an account is secure, since existing sessions may remain active. The four proposed controls are a uniform response to reset requests, bounded token validity, session revocation after confirmed reset, and a minimal audit record. Acceptance testing should cover expiry boundaries, token reuse, refresh-token rotation, and whether old and new authentication paths overlap during migration. Audit events are limited to a small application-owned vocabulary, and tokens, passwords and authorization headers are excluded from logs.

Re-Saving a JPG at the Same Quality Barely Hurts It, Tests Show

A developer tested how repeated saves affect image quality by re-encoding a single test photo seven times and measuring pixel-level error after each generation. Re-saving at the same quality with the same codec produced almost no additional damage, with error rising only marginally over seven saves. Switching between JPEG and WebP caused a noticeable jump in error after just one swap, and the error kept climbing slowly with each further switch. The author attributes the difference to how each codec re-quantizes the image, so changing formats compounds losses. The findings suggest that the greater risk comes from changing codecs or letting quality settings drop during re-saves, rather than from simply reopening and saving a file.

Guide Matches Local LLMs to RAM Sizes: Best Ollama Picks for 4GB to 32GB in 2026

A new guide ranks the best locally run language models for machines with 4GB, 8GB, 16GB, and 32GB of RAM, based on the authors' testing on real hardware. It explains that models quantized to Q4 need roughly 0.6 to 0.7 GB per billion parameters, plus 1 to 3 GB of overhead for context and the KV cache. The authors argue that newer small models have improved enough that older recommendations are now misleading. For GPU users, the guide notes that available VRAM is the main constraint and that GPU offloading generates text several times faster than CPU inference. Each RAM tier is paired with specific model suggestions and expected speeds, such as roughly 6 to 10 tokens per second on CPU for a 14B model.

Memotong video dari transkripnya dengan ffmpeg, dan jebakan jeda yang tersisa

Penyuntingan video berbasis teks sebenarnya hanyalah daftar rentang waktu per kata dan satu perintah ffmpeg, di mana kata yang dihapus menjadi potongan sementara sisanya dipertahankan. Penulis menjalankan pipeline ini dengan Python pada klip wawancara NASA bersama astronaut Christina Koch berdurasi 37,3 detik, yang transkripnya berisi 119 kata. Audio cukup diekstrak untuk dikirim ke transkriber agar biaya lebih rendah, dan waktu potong dibulatkan ke batas frame 30 fps agar sesuai dengan yang dipotong ffmpeg. Hasil suntingan menghasilkan empat rentang, termasuk satu rentang pendek 134 ms yang memuat kilasan tengah kalimat yang seharusnya sudah dihapus. Masalah itu muncul karena jeda di antara kata-kata terhapus ikut dipertahankan, sehingga perbaikannya harus menentukan kepemilikan jeda tersebut.

Voters Push Back on Data Centers, Shaking Silicon Valley's Election Hopes

Silicon Valley appears to have taken a political setback as public sentiment against data centers becomes a factor in elections. The Atlantic reports that many Americans hold unfavorable views toward these facilities. The source material provided is limited to a headline, a link, and a small amount of Hacker News engagement, so specific details such as polling figures, named candidates, and locations could not be confirmed. The story is framed around the tech industry's reduced political standing amid local concerns over data center development.

Docraft Chrome Extension Converts Pasted AI Responses to Word, PDF or HTML Locally

Docraft is a Chrome extension that turns pasted AI chatbot responses into editable Word files, styled PDFs, or standalone HTML pages. It processes everything inside the browser, with no server, account, analytics, or AI API involved. Users can paste text into a compact popup for quick export or open a full workspace with an editor, live preview, and document settings. The tool supports Markdown-aware formatting, light and dark documents, batch ZIP exports, and KaTeX-rendered math in PDFs. It is open source under the MIT license and currently installs from source code, and PDF output may not fully render CJK or complex scripts.

How to follow news sites without RSS feeds by tracking their section pages

Many large publishers, such as Nikkei, offer no RSS feed, so standard feed readers report that no feed was found. Their section pages still list new articles, and a reader can monitor those pages and extract the article links. In the osFoundry News screen, a user adds the site address and, when no feed exists, is offered the publisher's sections drawn from its own sitemap. Selected sections become individual sources with their own check schedules, each requiring a single request to the section page. Article-shaped links are filtered from the page, and headlines and times are taken from it, with short summaries read from each article's description tag. Sites that build their lists with JavaScript or block automated requests cannot be followed this way.

Sinh viên phát triển website chia sẻ hình nền máy tính 4K miễn phí

Một sinh viên đã xây dựng website wallpaperPC trong quá trình hoàn thành đồ án môn học. Trang web cung cấp hàng nghìn hình nền máy tính chất lượng 4K và cho phép tải xuống miễn phí. Giao diện được tối ưu để hiển thị tốt trên nhiều thiết bị khác nhau, đồng thời tốc độ tải trang được chú trọng. Người dùng có thể truy cập trực tiếp để trải nghiệm dịch vụ.

Upset J&K National Conference MLAs to Meet Farooq and Omar Abdullah Today

A key meeting is scheduled today between Farooq Abdullah, Omar Abdullah and dissatisfied National Conference MLAs in Jammu and Kashmir. At least 18 of the party's 41 MLAs are pressing for sweeping reforms in how the government and organisation function. They are also seeking greater accountability from leadership. The MLAs have additionally demanded the removal of a close aide of Omar Abdullah. The development points to growing internal discontent within the ruling party.

Q-comm eyes a bigger festive basket this Diwali

Auto dealers see clouds beyond festive cheer

Zomato founder Deepinder Goyal responds to doctors' criticism of Temple wearable

Deepinder Goyal has replied to doctors who have voiced objections to Temple, his wearable health-tech device. He rejected the suggestion that the product is a scam, pointing to the substantial money invested in it. Goyal also noted that customers can return the device within 14 days if they are unsatisfied. He said scientific validation of the device's claims would be published at a later date. He concluded by expressing confidence in the innovative features Temple offers to the wellness market.

Delhi police traffic advisory for Diwali season

Jyot Singh becomes first Sikh elected to Georgia State Assembly

Jyot Singh has become the first Sikh to be elected to the Georgia State Assembly in the United States. His election marks a historic milestone for the Sikh community in the state. The article reports on the significance of his win for representation of minority communities. Further details about the election and his plans in office are covered in the original report.