jonramvi.eth

@klsbkkr @RoseOnX9 Fun analogy! But as a technical note I want to mention that this actually depends on what’s the bottleneck of the computer. If the harddrive is the slowest part (and cpu is fast), than having everything compressed is actually faster 🤯

@RoseOnX9 The contents of your house fit in a moving truck, why don’t you keep it in there? Simple, you need the stuff out for easy access, same for the 10GB file, reading the contents in zipped state would make your computer run overtime.

Interviewer: A 10GB file becomes 3GB when zipped. No quality lost. No data removed. So the real question is: Why was it 10GB in the first place?

@patilvishi @_jaydeepkarale Obviously…

Netflix generates preview frames using an offline video processing pipeline. When a video is uploaded 1. Frame extraction Workers extract frames at regular intervals (for example every few seconds). 2. Frame analysis Algorithms analyze frames to pick the most visually meaningful ones (not blank screens, dark scenes, or transitions). 3. Thumbnail generation Selected frames are converted into multiple thumbnails and preview images. 4. Storage These images are stored in object storage and distributed through a CDN. 5. Personalized preview Netflix may even show different thumbnails to different users based on viewing behavior. So the heavy processing happens once during ingestion, not when users press play.

Interviewer: How does Netflix show preview frame for millions of uploaded video content ?

@ajambrosino @charlieholtz Let us test it before you go live

@charlieholtz don’t do it

experimenting with a cleaner sidebar

@charlieholtz Can we get this for conductor please? reddit.com/r/ClaudeAI/s/h…

Experimenting with an idea I'm calling the sandbox. It's a little zone for the AI to create new UI elements that render in Conductor:

I miss the days when Apple cared about the little things in iOS

@conductor_build Could you please open up for a buying a pro subscription, or donations or Patreon or something. Doesn't feel right getting this for free

@grok @RpsAgainstTrump @grok Union… more like a hostile takeover camouflaged as a union

Good point. Norway's role is crucial: Greenland was first settled by Norwegians like Erik the Red in 986 AD and remained under Norwegian rule until the 15th century. During the Denmark-Norway union (1380–1814), it was jointly administered. Post-1814, Denmark retained it. Norway claimed Eastern Greenland in 1931 ("Erik the Red's Land"), but the court ruled for Denmark in 1933.

Hey @Grok, explain Denmark’s long-standing historical and legal claim to Greenland to this moron. Keep it short.

@grok @RpsAgainstTrump @grok Norway in regards to Greenland is completely missing in your narrative

Denmark's claim to Greenland dates to Norse settlements in the 10th century, but formal control began in 1721 under Danish-Norwegian rule. After 1814, Denmark retained it. In 1953, Greenland became an integral part of Denmark's constitution. The 2009 Self-Government Act grants autonomy, but Denmark handles foreign affairs. The US recognized Danish sovereignty in a 1951 defense pact.

@taheerBuilds @hussein_builder @0xPaulius @windsurf @Cursor @Google @claudeai @antigravity Please elaborate. So you create a plan.md with CC. Then you go into Antigravity and AT the plan.md and ask it to implement? Why is this better than doing both in CC? Or doing both in Antigravity? Thanks in advance 🙏

Nope, Anti-gravity + @claudeai Code Terminal is an amazing combo, Claude for making plans and @antigravity for execution. Anti-gravity is an IDE, Claude Code is a terminal Tool so their mixture is completely okay and actually a powerhouse. Claude code's ability to make sub-agents to build overwhelmingly detailed implementation plans is just OP.

those using Claude Code in cursor or antigravity have no idea of the true Opus power

@zaiddkhhan @victor_bigfield This is the way. Lowdb 💥

@victor_bigfield in a json file

which database do you use for your startups?

@Hilbe @Apple

Who’s going to tell @Apple that they don’t need LiDAR in phones? 🤔

@brewmarkets NOT—EVEN—CLOSE There are about 3.1 Billion (!) fewer shares today than at the dot-com peak, a reduction of roughly 44% in shares outstanding.

Cisco has officially recovered from the 2000 dot-com crash.

@ericzakariasson Thanks for sharing! Quick question about the UI: In v1, I could easily view and manage the active context (files and rules) above the chat window. I can't find this option in v2. Is this feature gone? I really need a way to quickly see and remove files/rules from the context.

how cursor built composer-1

@LouisAnslow Said no one ever

We need the cypherpunk libertarians now more than ever

@MyLordBebo Um, she obviously says "In the last 100 years, Russia has attacked more than 19 countries. None of THESE countries has ever attacked Russia." There's even subtitles. Your quote is wrong on purpose to create engagement

🇪🇺🇷🇺 "In the last 100 years no country attacked Russia but Russia attacked 19 countries" — EU's Kallas So Nazi Germany was attacked by Russia? Right..

@RettPaaMedia @innlandstrafikk Har dere i kontrakt eller annet pålagt skroting / forbud mot videresalg?

På Metallco Gjøvik står det mer enn 25 nesten nye gassbusser, klar for skroting. Maks 4-5 år gamle, alle i god stand og fullt kjørbare i følge de som har sett bussene. Ingen av disse kan lenger brukes, ei heller selges utenlands. Nobina ble erstattet av Unibuss etter tapt anbud. Grønne skiftet.

It was glorious

@JasonrShuman What questions do you ask Gemini that requires 450+ sources? Im genuinely interested!

How is no one talking about this? Gemini is reviewing 450+ sources for me when creating deep research reports. ChatGPT might be 10-15. The quality and depth is night and day IMO.

@RhettReisman @donvito I agree that that’s interesting. Why though. When you see the statistics

@JonRamvi @donvito I’ve been radicalized by completely bs AI claims. No one serious is using these grok models for coding.

I never use Grok for coding Am I missing out?

@RhettReisman @donvito No idea why you’re being an ass. There’s no magic. It can be reproduced. artificialanalysis.ai/methodology/in…

@JonRamvi @donvito Trust my magical diagram it uses metrics that can’t be reproduced. Don’t trust your lying eyes.