If you’re interested in this, the term you’re looking for is punycode

it’s no kagi, but its ok

thank you! this needed said.

• This post is a bit critical of a small well-intentioned project, so I felt obliged to email the maintainer to discuss it before posting it online. I didn’t hear back.

i used to watch the dev on mastodon, they seemed pretty radicalized on killing AI, and anyone who uses it (kidding!!) i’m not even surprised you didn’t hear back

great take on the software, and as far as i can tell, playwright still works/completes the unit of work. at scale anubis still seems to work if you have popular content, but does hasnt stopped me using claude code + virtual browsers

im not actively testing it though. im probably very wrong about a few things, but i know anubis isn’t hindering my personal scraping, it does fuck up perplexity and chatgpt bots, which is fun to see.

good luck Blue team!

discourse > discord

A heatsink works by increasing surface area to dissipate heat through convection (air moving past it) and radiation (infrared energy). The key principle is:

Heat dissipation ≈ Surface Area × Temperature Difference × Heat Transfer Coefficient

but, for me, math is hard, so…

Why You Can Cut Fins Off (Usually)

1. Diminishing Returns: Each fin you add provides less cooling benefit than the previous one because:

• Inner fins get less airflow (they’re shielded by outer fins)
• Heat has to conduct through more material to reach outer fins
• The temperature gradient decreases as you move away from the heat source

1. The Rough Rule: A heatsink typically operates with 30-50% margin. So if it’s rated for 5W and your RPi CPU draws 3W, you have room to lose some fins.
2. Fin Efficiency: There’s actually a mathematical concept called “fin efficiency” - fins that are too long or too closely spaced become ineffective. The last row of fins might only contribute 10-15% of total cooling.

Napkin Math Example

Say you have a heatsink with 8 fins:

• First 4 fins: ~60% of cooling
• Next 2 fins: ~25% of cooling
• Last 2 fins: ~15% of cooling

Cutting off one row (2 fins) loses maybe 10-12% of cooling capacity. If your CPU was running at 65°C with the full heatsink, it might now run at 68-70°C - usually fine since RPi CPUs throttle around 80-85°C.

When You Can’t Cut Fins

• If the heatsink was already barely adequate
• If you’re overclocking
• If ambient temps are already high
• If there’s no airflow

The real question to ask: “What’s my current CPU temp under load?” If it’s 60°C, cut away. If it’s 78°C, maybe find a different heatsink.​​​​​​​​​​​​​​​​

reductive shit.

nice try… onion…

:y priest_corpse | :put graveyard | :%s/dignity/profit/g | :!extract organs | :s/quality/quantity/g | :%s/burial/assembly_line/g

made me think of the game graveyard keeper

this shows a fundamental misunderstanding of how serverless can be used, and modern techniques. serverless is absolutely a valid choice if you architect for it. nerdbait article

This project should be significantly easier to for folks if you add claude code to your stack. cc is an excellent admin, and server manager.

yep

1password

Internet Protocol.

ipv4 remains dominant.

tcp and ip merged in like 1973, and it lived in labs till 82 or 83. after that its been 40 years of nearly perfect ip spec

the meta youve heard js harmful.

“Brandish or draw a gun…better be prepared to kill with it.” — Normative, but the law is dead serious.

Illinois doesn’t have a standalone “brandishing” statute; threatening display can be charged as assault (often aggravated assault when a deadly weapon is involved). Separately, deadly force is only justified if you reasonably believe it’s necessary to prevent imminent death/serious bodily harm (or a forcible felony). Drawing in a way that threatens without lawful justification can be a crime.

Read up on shodan.io. bot networks and scrapers can use the database as a seed to find open ports.

The cli massscan can (under reasonable conditions) scan the the entire ipv4 address space for a single port in 3 minutes. It would take an estimated 74 years for massscan to scan all 64k ports for the entire ipv4 network.

So, using a seed like shodan, can compliment scanners/scrapers to isolate ip addresses to further recon.

I honestly don’t know if this helps your question, I don’t actually know how services in general deal with nonstandard ports, but I’ve written a lot of scanning agents (not ai, old school agents) to recon for red/blue teams. I never started with raw internet guesses, I always used a seed. Shodan, or other scan results.

Wow, you really told me. I hope you feel.

I’ve learned my lesson. Good job.

Look out, I’m about to mention AI.

I had most arr things I wanted setup already. I have a putio stack mostly.

I decided to refactor the setup a while ago, shit was always breaking, and I just didn’t want to put time into it.

So I launched Claude code, started a server-manager repo, and about 3 hours later, I had a EXCELLENT setup, all my fave release groups on priority, resolutions I want, it found ally Plex media, stuck it in sonarr, reccomended I start a trakt account, which I did, and now every upcoming scifi, fantasy, horror, action flick gets picked up automatically, it created its own putio uploader, and downloader, I bought filebot on it’s reccomendation, and now all my media is renamed perfectly, I signed up for subs, I have subtitles everywhere. My audiobook library was imported, and it figured out what audiobooks I buy and what I don’t, gave me a stern talking to about piracy, then I launched opencode and used grok4 to process my audible books to remove drm. (Grok really doesn’t say no to many requests. However fuck Elon, so i don’t use it more often.) Claude setup a docker environment with qbitorrent, with a deadmans switch, it will only connect via my VPN. So I could get rid of putio, which I won’t.

I asked for monitoring and it spun up uptimekuma. In the middle of the project we decided to use coolify toanage everything. Again, it set the whole f’n thing up. This was the only thing I had to finalize, UNTIL I found out there’s an API, so Claude read the API docs and now runs my coolify platform.

Honestly the afternoon was fun as hell, claude-code just set everything up, I had to guide/steer it some, but it was able to read documentation on its own and tighten up my setup. It even optimized my docker to properly use my gpus. Now I have comfyui, ollama +gptoss, also running on my network with very little effort.

I have a max subscription, and after paying a few bills without it, I won’t go back. This setup burned a lot of tokens. It easily would’ve been a few hundred if I’d payed API rates. I think however a pro account would have enough tokens in a day to do something similar. You don’t even need Opus, Sonnet did fine.

So, if your stuck, get you a Claude pro subscription, hook up Claude code, and see if it can help you.

I use Claude almost all day, so this was pretty easy, but I cannot stress enough how much I dislike administering my media setup. Now I have a Claude custodian.

Software typically has features because people want them, so unnecessary is utterly relative

i use a custom n8n routine to gather and vet news. im just trading one ai for one I make the mistakes with

I started with this in 2021 https://mgdm.net/weblog/csp-logging-with-nginx/

At some point in 2024 i got Claude to take over this management. Now i have an agent that maintains this workflow, and all csp reports are in nginx/json