Trailblazing Braille Taser

Sounds like you were a 1337 H4X0R back in the day

Did you tell him about fuzzers yet?

No, it’s JS

I don’t do much frontend work these days, but years ago, it felt like the defining feature of Bootstrap was the 12-column layout. Sure, it had fun buttons and other components, but the ability to trivially define multi-column layout without ripping your hair out was its raison d’être.

Now that we have flexbox, I’m not sure anyone needs Bootstrap.

Someone called me a cat dad and it sounded so weird to me. I like to think that my cats view me as “big friend”. I’m no father, more of a guardian who wants what’s best for my silly creatures and would die for them if it came down to it, but certainly not a father, no siree.

Well they’re not metal detectors, now are they? (This fact brought to you by Detectorists on BBC Four.)

I’m no AI proponent, but I’ll only believe that LLMs are causing this psychosis when I see a randomized controlled trial. Until then, it seems far more plausible that people experiencing delusions gravitate towards LLMs.

Whenever I try building something with wood, I get so frustrated that it’s not version controlled. In software, I can fearlessly try dumb stuff because I can just roll it back if it didn’t work.

Humans weren’t “meant” for anything.

Hi, it’s me, the creator. Your purpose is to create entertaining content, but your output has been slipping lately. You live in a simulation created by my multiversal corporation, which didn’t meet growth expectations this cycle, so we’ll be making some cuts to your simulation’s fidelity — just a few fingers and toes for now. Try refocusing your civilization on pumping out Boss Baby movies (they’re very popular here!) and we can talk again in 172 of your years.

Man Bun™ was created when a regular rabbit ate a radioactive turd. Now he can anthropomorphize at will, leap tall buildings in a single sproing, be super jacked, and destroy his enemies with Mach-5 poops.

Would you have preferred an awk script?

Yeah, exactly. I don’t get it.

You can’t do incremental builds with just. I don’t see how it’s a replacement?

You can also “simply” raw-dog Wireguard. It’s built into the Linux kernel, so you barely have to install anything besides the userspace tools.

Basically, I objected to being reliant on the generosity of a for-profit company. “We do these things not because they are easy, but because we thought they would be easy.”

This is a rough sketch:

• Create a Linux server. It can even be a VM/container if you get the networking right.
• Create a Wireguard interface and pick a private IP address subnet that won’t conflict with your home subnet: https://www.wireguard.com/quickstart/
• Define PostUp and PostDown rules in your Wireguard config that modify iptables to masquerade traffic from the Wireguard subnet
• Also set net.ipv4.ip_forward=1 with sysctl. (There is probably an IPv6 equivalent but I live in the past.)
• Generate keypairs and configs for each device you want to use
• Set up dynamic DNS, e.g. https://freedns.afraid.org/
• Forward UDP port 51820 to your server
• Install the Wireguard app on your client devices. If Linux, you can just write a client config containing the necessary magic words and start it up with wg-quick.

Boom. Tailscale’d.

I’m sure I’ve forgotten some steps. I have some janky automation that’s broken in a new way every time I try to use it.

I have no opinion of them, but I’m curious why advertising would imply untrustworthiness. Are you saying they’re too eager or something? Spending money on ads is also consistent with a company that’s making money by charging for a service — I’d be more suspicious of free VPNs.

Given the warning about capitalization, the best possible case is that they’re using ast.literal_eval() rather than throwing untrusted input into eval().

Err, I guess they might be comparing strings to ‘True’ and are choosing to be really strict about capitalization for some reason.

I am compelled to optimize this code:

pass

My knife is printf, but my scalpel is gdb.

I honestly think autocorrecting your scripts would do more harm than good. ShellCheck tells you about potential issues, but It’s up to you to determine the correct behavior.

For example, how could it know whether cat $foo should be cat "$foo", or whether the script actually relies on word splitting? It’s possible that $foo intentionally contains multiple paths.

Maybe there are autofixable errors I’m not thinking of.

FYI, it’s possible to gradually adopt ShellCheck by setting --severity=error and working your way down to warnings and so on. Alternatively, you can add one-off #shellcheck ignore SC1234 comments before offending lines to silence warnings.

PSA: Run ShellCheck on your shell scripts. It turns up a shocking number of programming errors. https://www.shellcheck.net/