e12e

Ah. You're assuming we're not living in a simulation?

I'm happy you got better - but isn't healthy diet, moderate regular exercise and good sleep hygiene staples of traditional medicine?

Sounds like:

https://www.ironlace.com/

And Google?

Wiping out a VM, server or workstation should not really be a problem - just restore from backup.

Silently corrupting files, that goes undiscovered until after backup window closes, and data exfiltration are the immediate, serious risks.

Wow, listening to the article [a], with the flat, tinny AI voice - it's hard to shake the feeling that this is satire.

And in the YouTube video, I think I glimpse the AI delegating and assigning work to the user (hey, you should review this file/presentation) - via Teams or some other chat interface that probably grabs your attention with an obnoxious ping!

Ironically, the vision they seem to be peddling looks quite close to what NocoBase[n] is selling.

Key difference being that NocoBase at least try to increase human agency.

They both suffer a bit from 4gl flaws; how do you version your code, how do you move it to another system, eg when changing employer.

Like, how do bring your customized Chad - senior assistant - with you to Google workspace or fastmail?

For NocoBase AFAIK they have modules, but no real versioning - so I don't see how it is a good fit for developers.

I just recently came across it as I was looking for a ticketing system - and no, I don't want to build my own in a 4gl system if I can help it. But holy shit would I be more inclined to let our org users play with NocoBase rather than set things on fire with chaos agents running on top of OpenClaw!

PS: We looked at libreDesk (too simple, missing things like merging issues) - looks like we'll go for FreeScout.

I'm kind of sad that Trac/Apache bloodhound died (and I think died, not just became feature complete) - they worked pretty well for email first support tickets.

[a] https://www.microsoft.com/en-us/microsoft-365/blog/2026/06/0...

[n] https://www.nocobase.com/

I read the readme that way too - a table with included utils with conflict status, and a list of intenationally excluded utils.

But the rest are in there:

https://github.com/microsoft/coreutils/blob/3fa7aaf832ffc81d...

Are you aware of:

https://docs.racket-lang.org/sweet/index.html

How long did it take from the first macro virus until the industry accepted that "we can't have nice things (at this cost to security)" - macros were defaulted to off everywhere?

How long until the industry accept the risk LLMs pose with "prompt injection"?

The css zengarden works fine on phones?

https://csszengarden.com/

Granted, then you're talking 2003.

The Ars Technica submission might be better (I've not read TFA):

https://news.ycombinator.com/item?id=48309492

They basically trained a neural network on the data they got from the SSD - and recorded data with other websites open in different tabs or even different browsers.

They could then guess/detect other open sites.

I presume, if they'd trained/recorded - they might detect other software as well.

But right now, they demonstrated (on MacOS) that if you open the exploit in a browser - they can look at SSD activity and tell you have website x, y and z open.

Might let you target users of a certain bank, child porn, regular porn, shopping sites... Mostly imagination that sets the limit.

... And looks like cosmopolitan libc wraps landlock for unveil, in addition to implement pledge.

Apparently someone tried wrapping landlock in unveil:

https://clehaxze.tw/gemlog/2022/04-02-landlock-unveil-experi...

https://github.com/marty1885/landlock-unveil

From above your quote:

> The only officially-supported operating system is OpenBSD, as this has considerable security features.

And below your quote:

> This is possible (I think?) with FreeBSD's Capsicum, but Linux's security facilities are a mess, and will take an expert hand to properly secure.

It is portable in the sense that it compiles and runs, not in the sense that it has the same security features.

I'd love to see pledge/unveil on (upstream) Linux - but I'm not holding my breath.

The article is a little one sided, as it doesn't touch upon MinID which is a government ID service, and Idporten which is an authentication service that allows use of different EIDs, like MinID and BankID.

MinID is only considered "secure" while BankID is considered "highly secure"; as the linked pdf report (on Norwegian) states - in Norway, due to the popularity/market dominance of BankID - a lot of the logins are "highly secure" - while in Sweden their (different, but with same name) BankID is only "secure" - and most services require only "secure" login.

In Norway there are AFAIK public services that require "highly secure" login - and there the public issued MinID isn't enough.

If 2fa for MinID is improved - I think it would easily be upgraded to "highly secure" (most other details are similar to BankID). That should take care of public services.

Private services that do not cater to the public good - would still need a portal similar to (or be granted use of) Idporten.

So I think catastrophe is a little hyperbolic - but the current path of BankID dominance isn't good.

Ed: I see the hn title is editorialized - TFA has a more balanced title.

Ed2: From the podcast - BankID might get downgraded to "secure" because of how 2fa is handled - so it's not only MinID that might need some adjustments.

I like to think of it like the difference between dropping a ball on a roulette wheel (get one random number/sequence of repeated) - vs dropping a ball on a carved topographic map, where valleys guide the ball to a particular outcome.

If you can stand a little AI expansion - here are a few points Gemini came up with - I think the idea has some merit:

https://g.co/gemini/share/b5b97867eeb1

(Maybe the better analogy is roulette vs pinball machine)

Ah, my bad. Wishful thinking going from GitHub to rails site.

Rails minor versions are a bit odd imnho - but you're right 7.1 isn't under support anymore.

They seem to be on 7.2, which will get security patches until August this year:

https://github.com/postalserver/postal/blob/8ef89606bc34146f...

https://rubyonrails.org/maintenance

Not great, but better than 6.x.

On a side note, for other helix users, I found two approaches to improve repl interaction from helix:

https://github.com/a3lem/replink

https://github.com/waddie/nrepl.hx

How so? I don't think anyone would consider Smalltalk to be homoiconic? But would consider it to support a repl-like development process?

Smalltalk has system images - which AFAIK clojure lacks (as does python, ruby).

I wonder if it would be possible to pair python ZODB with storing python code alongside the pickled objects... And effectively create an unholy image-like workflow with IPython and ZODB?

But at any rate, I was more curious about how you mix repl, clojure and LLMs in practice?

Curious what that would look like - and if javascript, ruby would benefit equally?

How do you work with LLM and repl?

> Languages with a single way to do things benefit the most: Ruby

I love ruby - but surely it's closer to Perl and "There's more than one way to do it" - than python which generally strives for "There should be one-- and preferably only one --obvious way to do it."?

https://legacy.python.org/dev/peps/pep-0020/

That said, I appreciate a trivial rust program that does a simple job - as someone not quite having taken the time to learn rust yet?).

I would probably not bother with the built-in cookies - if anything - have an option to "seed" ~/.config/friend/fortunes.txt or whatever is the proper xdg path.

https://docs.rs/crate/friend/0.1.0/source/src/main.rs

Seems overly complicated if the goal was a "simpler fortune" - a self-contained shell-script is perfectly capable of echoing random lines (I should know, I had one for generating email signatures).

On the other hand, if all one wants is a set of fortune cookies - there's no need for the new client - fortune-mod exists:

https://www.shlomifish.org/open-source/projects/fortune-mod/

Odd, I'd imagine Wikisource (in many/all languages) would be part of training data for all LLMs with SOTA ambition?

https://no.wikisource.org/wiki/De_knyttede_n%C3%A6ver

I was curious if the basic ai chatbots behaved differently - been a while since I tried the same prompt across models..:

Apparently Gemini is the late night news anchor:

https://g.co/gemini/share/3504289b8dc8

Chatgpt the art critic:

https://chatgpt.com/share/6a11e80e-523c-83ea-9a1b-03329f860c...

And grok somewhat of an apologist:

https://grok.com/share/c2hhcmQtNA_4071d2b9-ae39-43cc-abf1-cd...

Any reason this can't be compiled to wasm and run in browsers that way? Networking (protocol) issues?

Ed: I guess you'd need a way to run wasm in wasm, and a way to author wasm contracts - and there's not yet a wasm-hosted rust compiler?

Ed2: I'm not up to date on the state of self-hosted wasm compiler/languages... But I did come across:

https://github.com/remko/waforth

I guess it's time to dig out the thesis[t] again and look at movable code and p2p classifications again ...

[t] https://ics.uci.edu/~fielding/pubs/dissertation/net_arch_sty...

I'd argue the LLMs certainly distribute copyright material. That's why it can do things like:

https://g.co/gemini/share/20843b4609d9

Now, you could argue quotes are fair use - but can you argue the material isn't part of the LLM?

Interesting project - but you need to add some information on where the data goes. As far as I can tell, code goes to some upstream ai provider (for installing, for analyzing).

Telemetry goes to some provider or local hosted solution? And then to your upstream ai provider for analysis?

The catch is that if you don't know software development, you probably don't know what you need...