That’s what I meant. It’s a standard, and the sunk cost and network effect make it practical. Same as HTTP and SSL compared to SCTP and IPSec. Isn’t it sad when you see web devs preferring native apps even more than the general public?

Packaging is very practical, but that’s a boring dead end that unfortunately lasted 20+ years. I mentioned Triton in the context of Sun folks who have done containers better and earlier, yet failed to market because Sun went bankrupt. Your long list that was packaged is nothing that can’t be developed in-house better with far less complexity. Triton provides no benefit to you, but who knows maybe you’ll need dtrace one day.

Unikernels and Xen allow function as a service (I assume you don’t need that as well). FaaS is the future, as it’s a progression of the micro services trend. Startup time and hardening are not of interest to you as well, so no reason to switch.

I’m not completely against packaging. It’s great for open source desktop apps. When it’s out of sight away from the user, it turns into a boring game.

Triton and SmartOS are still Sun Microsystems people. Solaris zones preceded Linux containers by almost a decade. They lost the popularity contest, so they have to provide compatibility with kube and docker. I think infrastructure containers are zones with extra whistles.

For even better performance, I’d go with Xen and Linux unikernel. The startup time can be just 1s with such minimum overhead. It also greatly reduces the attack surface. I highly doubt you’ll be allowed for such a drastic change though.

So that’s how AI ended up killing

No. I didn’t mean to attack you in particular in any way. I apologize if it came off as such. I just dislike the blind copying of what Google and Facebook do. Docker is another atrocity that everyone seems to feel obligated to use.

Heroku supports moderately large amounts of requests. It’s less expensive than having a proper sysops team in most cases.

With 1,400 it’s probably worth it to move away from AWS. Something like self-hosted Triton (descendant of Solaris) cluster would be far more elegant than kube and lxc.

Something like heroku is better for out of the box stuff like logging and autoscaling. Some companies like banks have to have their own data center. But they should write their own “tools”.

Must be fun looking at 10 pods and pretending to be in control of Google search by proxy. Autists have a peculiar way of day dreaming, as they’re extremely limited in imagination. They always seek artificial complications to cover up the fact that what they’re doing is actually not far from trivial, and without gatekeeping a school kid would be capable of doing.

So? You’re limited by disk space?

Fair. I hate kube though. Most companies run just 10 pods because they cargo cult google. The complexity of it is completely unjustified

Why is it so funny to see issues with a terminal running in 4k? It never crossed my mind that some folks do that.

Konsole has no tabs?

I actually like Zig. I wish something like CUDA was available in it.

Thanks, this clears things up. I didn’t know what exactly was making print IO slow.

I don’t use any complex TUIs. Pretty much everything is CLI or GUI. Which TUIs did you have in mind that were slow?

I’d like to test this soon. I’ll look for a modern TUI framework.

I’ve never seen a slow terminal emulator. Most terminals have tabs and splits. Never experienced compatibility issues. Don’t care about Zig at all.

Are these all the reasons? Another toy software written out of boredom.

I must be retarded. People are excited about a terminal emulator. Why?

Good, because it’s not of interest.

Even better for your career is being an actual researcher with a good grasp on heap grooming, MAC circumvention instead of spreading FUD over something that can easily be detected with burp.

Large gains were due to scaling the hardware, and data. The training algorithms didn’t change much, transformers allowed for higher parallelization. There are no signs of the process becoming self-improving. Agentic performance is horrible as you can see with Claude (15% of tasks successful).

What happens in the brain is a big mystery, and thus it cannot be mimicked. Biological neural networks do not exist, because the synaptic cleft is an artifact. The living neurons are round, and the axons are the result of dehydration with ethanol or xylene.

AGI or human level intelligence has a hardware problem. Fabs are not going to be autonomous within 20 years. Novel lithography and cleaning methods are difficult for large groups of humans. LLMs do not provide much assistance in semiconductor design. We are not even remotely close to manufacturing the infrastructure necessary to run human level intelligence software.

It’s so much easier to quit social media when you’re on meth.

LLMs are not programmed in a traditional way. The actual code is quite small. It mostly runs backprop, filters the data. It is already easily generated by LLMs.

Is it more effective than feeding them via TikTok, Twitter, Instagram?

Let me guess, you signed an NDA, and won’t tell anyone which brands had badly configured access control in their web apps?

Each red flag is okay, but all together is rather strange. It’s kinda classic to say that pajeets write shitty code.

Notice a few red flags. 1) they were contacted before the system was rolled out. Before anyone else could look for bugs 2) it is reported by techcrunch and is trending tech news 3) the exploits are rudimentary 90s era mistakes that even LLMs don’t make these days

So it’s likely that they paid McDonald’s India to pretend to have horrible practices. $240 is another tactic to appear good and trustworthy. That brings traffic to their blue team company site, effective advertising. Standard fakery that security faggots utilize to spread the FUD to create demand for their services.