lcnmrn

I'm pretty sure there are hundreds of KBs of metadata with each tweet. For comparison, each page on Subreply takes 5 KB of data transfer and server response is around 100ms.

There are 28 letters. With 3 initial same words you can cover up to 53.5% of the alphabet. With 4 initial same words you can cover up to 71.4%. There are only 5 vowels A, E, I, O and U which can be covered with the first two words. You can even add Y in the first two words played.

Or someone will actually build a nice platform for once, like, I don't know, maybe Subreply.

As a social network owner of Subreply.com, I get to compete with Elon. I validate accounts before they start posting. Create account, improve profile, get approval, start posting.

No need to ask user for payment. No spam detection is necessary. No checkmarks either since everyone is validated.

The next Google should be powered by AI/ML algorithms to go around SEO tricks. If lots of people pick a result from page 16 that result should be on the front page.

What would be the Python, server-side version of it?

Jump from 99 to 109 then let them break. Internal sites of universities and corporations need to work until they get fixed.

Build Subreply. Tell people. Marked as spam. Build Unfeeder. Tell people. Marked as spam. And so on... :)

Clipboard should be a private and secure enclave (encrypted) in the first place. Only the OS should store and retrieve data from it.

Can you assign defaults with make_dataclass() but no types?

NamedTuples work with and without type hints, but Dataclasses only work with type hints: https://subreply.com/reply/lsk

That's was the initial purpose of Subreply.com just for friends, but it evolved past that and now I have to review each new account. It's still great because everyone it's verified.

I wish there was a service to allow only real people to sign up on small sites / forums.

We need a street football (soccer) game and procedural generated stadiums, players, teams. OpenAI can contribute a top engine to an open source game.

As the name implies, it takes feeds and brings the best from them to surface. Every click on article links counts as vote from a unique IP address.

It's created by me. I enjoy the read view on Unfeeder for long articles. I don't longer use or need a feed reader since I created it.

Actually, the fact that is not the same as desktop browser.

Vivaldi is the best browser on Android, but Firefox is king on Win/Mac/Linux.

I scaled Subreply with the same amount of money I would pay for Netflix/Spotify.

# highlighting for all languages

include /usr/share/nano/*.nanorc

set autoindent

set minibar

set morespace

set historylog

set linenumbers

set softwrap

set tabsize 4

set tabstospaces

Now you got a full featured IDE.

WASM is to Node what C/C++ extensions are to Python.

Math.js does the same thing. Demo on lucianmarin.com/num for some math, works on the phone too.

Why not build a container with battery inside and solar cells on top? Deploy at the top of the cargo ship. Equip the cargo ship with an electric motor.

JPEG XL has a non backwards compatible mode too for highest quality/compression and the bitstream can be improved with better encoders in the future.

The only missing part is support in all browsers and operating systems.

Crazy idea: what if we can use mRNA or any other tech to make old blood young again by changing the chemical composition? The rest of the body follows along, it's proven.

At this point, Firefox should adopt Gemini to protect the open web.

Helium is a much better, faster alternative. It offers fallback to Selenium WebDriver.

https://github.com/mherrmann/selenium-python-helium

Yes, it doesn't yet render pages as good as other browsers, but we can tweak our code for it. 10x better performance and 100x less memory usage is a win.

An alternative to Electron should be built using Servo and Deno, both written in Rust.

Tech world needs a TSMC competitor.

I'm working on new features for https://subreply.com social network/forum.