perbu

Back in 2008-2009 I remember the Varnish project struggled with what looked very much like a memory leak. Because of the somewhat complex way memory was used, replacing the Glibc malloc with jemalloc was an immediate improvement and removed the leak-like behavior.

Markdown is the native formatting of large language models. As such, support for it will be everywhere in the coming years.

I suspect LLMs, not users, have been the requesting this feature at Apple.

I'll suspect it'll be more like the next little thing. Most of don't interact that much with structured data, so the applications will be very specific.

However, the algo-trading crowd, will likely be very interested in this. They deal with structured data all day and it would surprise me if most of them don't already have things like this working in their networks. They seem to be very secretive, though, so we're not gonna hear much.

Yeah, but aren't these bussed mostly subject to state policy and not federal policy?

TinyKVM is now integrated into Varnish Cache Deno has been ported to run inside it.

This is the magical thing that happens when AI research happens in the open. Deepseek published their model and their methodology and then the nice people at the University of Illinois are able to build on it.

When OpenAI was launched this is what I thought it was going to be like. Something, something for the betterment of man kind.

No, it isn't. This is Cloudflare passing exposing metadata when it really shouldn't. Having a configuration option or a origin response header akin to CloudflareCache: private or something is trivial for them to implement.

The same information would then be available in the timing, but given the distributed nature here, that would be a lot harder to pull off.

I think the question "What is Tiktoks speech?" was raised. And the answer, "the algorithm" didn't really strike home.

So I read it like they didn't interpret this as a free speech issue at all.

sudo also had the ominous warning, "This incident will be reported!" when you fail to authorize the request. ofc, it just just logged.

Thanks! Even if this is only the enthusiast crowd, they're typically ahead of the curve, by a year or two, so Intel can't affort to ignore this. At some point the there will be less willingness from Dell/HP/Lenovo/Fujitsu to buy a subpar product.

Both. They can make acetic acid from both sugar and ethanol.

Looks great. I'm abit worried about it being free, though. People that take on arduous tasks for free tend to get tired of it after some time.

Open source helps a lot here, though, so I'm not too worried. Source is here, fwiw: https://github.com/timfraedrich/OutRun

Edit: Looks like the project is abandoned. So, yeah, maybe it should have costed a couple of bucks.

It is pretty good. Most of the CPU is spent on crypto, which is what you'd expect. The overhead is low enough that I've had no problems having rather meager machines handling thousands of concurrent connections.

If you're having performance issues with TLS I would look at what sort of crypto you're using. At least for SSH, RSA is dog slow. It wouldn't surprise me if you can irk out quite a bit of performance by switching to ed25519.

I suspect if it works well 99% of the time, which is pretty good, they're about half way to their goals. Making it work well 99.9999% is probably a lot harder.

Well done.

The ease of use and high quality of the Go SSH libraries (golang.org/x/crypto/ssh) is a killer feature of Go, imho.

Also, there is a high level abstraction, github.com/gliderlabs/ssh, which makes it completely trivial to embed an ssh server into an application, giving you a nice way to inspect counters and flip feature flags and tuneables.

Is XDG_CONFIG_HOME Unix? Isn't it just some Linux convention?

There are quite a few CDNs based on Varnish out there. Most of them are private, though, so you don't really see them.

They forked Varnish (around 2.0, judging by the VCL syntax) quite early and I suspect there isn't much left of the original codebase.

I'm very impressed how they managed to retrofit multitenancy.

fwiw; I worked in a company that was kind of a Linux pioneer back in the 90s. We ran Linux for everyone. The accountant ran a VM for his accouting software, but except that everything was Linux. And once set up, it worked very well, for techies and non-technical alike.

If you get past installation and initial setup, using Linux in a desktop role isn't really challenging if you've got access to support.

Sure. A cloud provider can copy your data without you knowing, they might migrate a host from one cluster to another and copy the data along with it and I would find it unreasonable to be held accountable if they forgot to scrub the source.

It would still be my problem at the end of the day.

I assume it'll be a paid API so the "contract" is a lot more clear. Twitter never understood what to do with its API so pulling that particular rug makes sense.

But I too wouldn't use this. X is playing fast and loose with ... everything, so having a business rely on their product seems risky.

Shit might happen with the best testing, but with decent testing it would not be this serious.

Software doesn't have uptime guarantees. They might have time-to-fix on critical issues, though.

I assume this is gross negligence, which would leave them open to claims made through courts, though.

Just NFS, I assume. It doesn't take more.

It is hard to imagine a more opinionated system than GNU.

"Never attribute to malice that which is adequately explained by stupidity." (stupidity is the wrong term here, lazyness is more apt).

Firefox is a the one different browser with an entirely different pedigree. Assumptions made about _all– other browser will not necessarily hold true for Firefox.

I've had similar experiences. Switching to Safari seems to help when this happen. I don't expect they'll be doing this on purpose.

It could be that the encoded video block required by Firefox and not used that much, as such the cache hit rates are bad requiring trips back to an overloaded origin server.

Machine failures are few and far between these days. Over the last four years I've had a cluster of perhaps 10 machines. Not a single hardware failure.

Loads of software issues, of course.

I know this is just an anecdote, but I'm pretty certain reliability has increased by one or two orders of magnitude since the 90s.

I moved away from the Terminal because of lack of utf8 support, also I "needed" 24 bit color support - I was working on some silly 24b gradient progress bars at the time.

I've been a happy Kitty user for some years now. It requires a specific termcap, so foreign servers are a royal PITA, otherwise it is pretty nice.

The bootloader is read-only, the firmware it boots is updateable.

The bootloader allows you to update the device with new firmware. MCU are different than computers. And since you can't overwrite the bootloader, you can't brick the MCU. You can always just reset it.

fwiw; I've never bricked an MCU buy flashing something weird onto it. The hobbyist MCUs sold are typically quite easy to re-flash with new firmware.