MartSec

@lancedb @kellabyte This metric is cheating. If you are a table format you should compare yourself to delta or iceberg. Not to a file format that of course you need to rewrite everything.

Adding a column to a 10M row dataset: 𝗟𝗮𝗻𝗰𝗲 - 𝟭𝟯𝗺𝘀 ⚔️ Parquet - 520 seconds. Lance writes new columns as separate files alongside existing ones — the original data is never touched. We benchmarked Lance v2.2 across multimodal training workloads at TiB scale 🧵

@bakubaku1899 @3CatInfo Pel que vaig mirar per endesa, has d’esperar alguns anys a veure si l’@AEPD_es actua i si no, pots intentar fer una denúncia. Però que acabaran dient que prenien mesures molt bones i molt boniques i que nanai. Que pobrets

@3CatInfo Algun advocat sap si es pot i com s'ha de denunciar això? Les empreses no tenen cap tipus de responsabilitat per no protegir adequadament les nostres dades?

Booking alerta d'un ciberatac amb accés a dades personals de clients 3cat.cat/3catinfo/booki… 3cat.cat/3catinfo/booki…

@1Password @AdamMcG02752243 Hi, migrating to a new region is not very friendly… isn’t there a possibility you could improve it? Tbh, asking family nom technical family members to follow it is a huge friction. It’s like the contrary of your great user experience.

Hi Adam, thanks for reaching out. You can create an account on our .com, .ca, or. eu websites which will determine where your data is hosted, and the currency you're billed in. With the mention of USD, it sounds like an account was inadvertently created in the US region instead of Canada. Our apologies for any trouble here. You can find steps to change regions in our guide linked below. Our support team would be happy to help with any questions about this process. If you still haven't heard back from the team could you direct message us here with your email or ticket ID? support.1password.com/regions/

Hey @1Password — you advertised $5.24 CAD/month. After my trial you want $71.88 USD. That's nearly $100 CAD. Then you ignored two support emails. You're a Canadian company. Act like it. 🇨🇦 Bitwarden it is. 👋 #1Password

Here's my current workflow summary for spec-first agentic development in Rust (example from ring-buffer repo [1]) .. The order of evolution of artefacts both for new development and for updates is: spec.md → Quint spec → Rust code • spec.md (source of truth) - define or update the invariant in English + math notation • RingSPSC.qnt - translate the invariant into a machine-checkable Quint model, add tests, run 𝚚𝚞𝚒𝚗𝚝 𝚟𝚎𝚛𝚒𝚏𝚢 / 𝚚𝚞𝚒𝚗𝚝 𝚛𝚞𝚗 --𝚒𝚗𝚟𝚊𝚛𝚒𝚊𝚗𝚝 to validate the model • Rust implementation - generate/update three artifacts downstream and then proceed for the rest of the implementation: • 𝚒𝚗𝚟𝚊𝚛𝚒𝚊𝚗𝚝𝚜.𝚛𝚜 - 𝚍𝚎𝚋𝚞𝚐_𝚊𝚜𝚜𝚎𝚛𝚝! macros referencing spec IDs • 𝚙𝚛𝚘𝚙𝚎𝚛𝚝𝚢_𝚝𝚎𝚜𝚝𝚜.𝚛𝚜 - proptest strategies exercising the invariant • 𝚚𝚞𝚒𝚗𝚝_𝚖𝚋𝚝.𝚛𝚜 - model-based test driver replaying Quint traces against real code Main components of Quint spec: 1. Constants 2. State Variables 3. Pure helper functions 4. Invariants 5. Actions (State Machine) 6. Temporal Properties 7. Embedded tests For a complete sample for the ring-buffer implementation, refer to [3] - around 640 lines of clean unambiguous spec. Why spec-first matters in agentic development: • Catches design bugs cheaply - model checking explores the full state space in seconds. A wrong invariant or missing guard is found before writing any Rust • Generates test oracles - the Quint spec produces ITF traces that drive MBT tests, so the spec is the test suite • Constrains the LLM - when the agent generates Rust code, the spec IDs (INV-SEQ-01, etc.) anchor it to verified properties rather than hallucinated logic • Feedback loop - if 𝚚𝚞𝚒𝚗𝚝 𝚟𝚎𝚛𝚒𝚏𝚢 fails after a spec change, you fix the model before touching Rust. If MBT tests fail, the spec and implementation are out of sync So update the Quint spec first, verify it passes, then proceed to implementation. For details on the flow refer to [2]. [1] Github Repo: github.com/debasishg/ring… [2] Detailed development pipeline: github.com/debasishg/ring… [3] Sample Quint Spec: github.com/debasishg/ring…

Qwopus on a single RTX 3090. Claude Opus 4.6 reasoning distilled into Qwen 3.5 27B dense, running through Claude's own coding agent (claude code). 29-35 tok/s with thinking mode on. the jinja bug that kills thinking on base Qwen doesn't carry over. harness and model matched. the base model would pause mid task on Claude Code. just stop generating. that's why i ran it through OpenCode, which handles stalled states automatically. this distilled version doesn't stall. it waits for tool outputs, reads them, selfcorrects when something breaks, and keeps going. i gave it a benchmark analysis task. went 9 minutes autonomous. wrote a README nobody asked for. zero steering. video is 5x speed but fully uncut. if you have a 3090, you can run this right now. free. no API. no subscription. opus structured reasoning on localhost. octopus invaders is next. same prompt that base qwen passed in 13 minutes and hermes 4.3 failed on 2x the hardware. i want to see if the distillation changes the outcome or just the style. more data soon.

For those of you rediscovering TDD via agents - if you let your agent do TDD by writing unit tests and not developer tests, you will end up in a pickle.

@korewadiego És un telèfon android en veritat. Amb dues coses extres.

com envien les alertes perquè apareguen ací wtf

@pedropreciadon @SergioBleynat Esta pregunta se ha respondida por hechos consumados, como dices.

@SergioBleynat Ojo: "¿puede una IA decidir matar a alguien sin intervención humana?" no la está respondiendo ningún parlamento, ninguna corte, ningún tratado internacional. La está respondiendo el mercado."

1/Hoy a las 5:01pm en Washington se decidió algo que va a definir cómo se hacen las guerras por las próximas décadas. Y casi nadie lo está leyendo bien.

I es poden quedar curts. Ja ho deia el de google fa quasi 20 anys que més o menys sabien el que pensaves. Imagina’t ara.

Tota informació que ha tocat un mitjà digital ja l’heu de considerar compromesa. No hi ha altra. No hi ha volta enrere. Cap empresa vetllarà per la vostra informació de forma adecuada. Ho poden intentar però tard o d’hora cauran o ells o els seus proveïdors.

@senyorrossinyol @jordiganchitos AAAHHHHAVHAHAHHAHAHA si no et xerren en català. Què t’empatolles paleto

@jordiganchitos No, és un dels pocs llocs de Barcelona en el qual encara sents que ets a Catalunya. Us fa ràbia això, que ens estimem catalunya i el que és nostre

@Jjjjjjjjj13jjjj @Tinet_14 Un bar de tota la vida? Aquest? Crec que no eh. La pinta a tres pijos joves que han contractat cambrers a preus baratos i només apareixen a cobrar es veu just al entrar.

@Tinet_14 Ara que hem mig perdut el bar de tota la vida, anar a un bar de tota la vida s'ha posat de moda.

D'on ha sortit tot aquest boom amb el Bar Bocata aquest últim any? Tan bo és?

🧵La història de com un grup de hackers que intentava trobar maneres d'enganyar la verificació d'edat de Discord, va acabar descobrint una dinàmica de vigilància sobre les persones que faciliten una imatge al sistema de verificació Persona, finançada per Peter Thiel de Palantir.

Aquesta descoberta demostra que aquest tipus de sistemes de verificació d'edat son en realitat“ un cavall de Troia, per tal d'aprofundir en el despelgament de sistemes de vigilància massiva. FONT: therage.co/persona-age-ve…

@lalgorisme “Un Error” 😉😉💪💪

SEGUIM: Microsoft afirma que un error a Microsoft 365 Copilot ha provocat que l’assistent d’IA resumís correus electrònics confidencials des de finals de gener, eludint les polítiques de prevenció de pèrdua de dades. bleepingcomputer.com/news/microsoft…

Contrasta l'actitud del Parlament Europeu amb la IA vs la de l'administració de Catalunya, que fa un sí a tot alhora de donar accés a les dades de milions de persones a Copilot, la IA de Microsoft. Ningú al volant.

Insostenible

@brave @Pirat_Nation Yes. On mobile. And random video stops and brave freeze

@Pirat_Nation Is anyone having these issues in Brave? Seems like most replies here are saying no but if you are, lmk and I can alert the team. 🙂

YouTube is escalating its crackdown on ad blockers. Many users report that comment sections now appear as "turned off" on videos, and video descriptions are missing entirely. The issue stems from ad blocker detection; disabling the extension immediately restores both features. No reliable workaround exists yet beyond turning off ad blockers or subscribing to Premium.

💩🖕

@ProtonVPN @midudev It works at ISP level. Here court of law and rights means nothing if football is involved. Shitty country

@midudev x.com/ProtonVPN/stat…

Esto no es una victoria contra la piratería. Es una derrota para la neutralidad de Internet y la privacidad. Que LaLiga y Telefónica consigan que se obligue a NordVPN y ProtonVPN a bloquear IPs sin garantías es un precedente que debería preocuparnos a todos, incluso aunque no veas fútbol. Una VPN es una herramienta neutral. Igual que un navegador, una conexión a Internet o un cuchillo. Como no soy capaz de detener al delincuente, decido retirar los cuchillos de los supermercados. Y lo peor es que gente que, como yo, no consumimos fútbol de ninguna manera nos vemos igualmente afectados por este tipo de decisiones. El problema real no es técnico. La piratería no existe porque haya VPNs, Cloudflare o Internet. Existe porque el producto es caro, está fragmentado y es difícil de consumir legalmente. Si alguien no paga ahora, no pagará mañana. Un pirata no deja de serlo por castigo, deja de serlo cuando el producto y el modelo cambian. Spotify, Netflix y demás lo demostraron. Y luego está el papel de cierto periodismo deportivo. Medios como Marca actúan aquí como simples altavoces de comunicados corporativos, sin análisis técnico, sin contexto legal y sin el más mínimo escrúpulo sobre las consecuencias reales de lo que están amplificando. No informan, legitiman. Y cuando renuncias a cuestionar el poder para limitarte a reproducir su relato, dejas de hacer periodismo y pasas a formar parte del problema.

(2/10) You know dynamic pricing—think Ubers, flights, or concert tickets that surge based on supply and demand. “Surveillance pricing” takes this to a new level: using your data to set a “price for you” based on your predicted breaking point. This is, increasingly, everywhere.