apt-get

We found it pretty easy to build a little k8s controller for our own purposes to do this -- see https://news.ycombinator.com/item?id=48478994 . You probably don't need to implement this as a plugin or hook, pgdog supports dynamic reload of its configuration without dropping existing connections.

Although I'm the type to shy away from adding extra layers in my architectures when I can help it, pgdog has been an absolute breeze to use :)

basic auth is secure, if used in combination with TLS.

Considering "I Can Has Cheezburger" and that kind of lolspeak is nearly 20 years old now, I'm not surprised to see some people unfamiliar with it nowadays :-)

We successfully did this with pgdog at $JOB using our own "controller" -- the same service that handles deploying new instances of our application (instancing an argoCD Application that fires Crossplane DB creation, making new Deployments of bricks, etc) will also, at the end of that process, scan the cluster for Database CRDs, use those to generate a new pgdog.toml + users.toml, update the Secrets in the cluster, enable maintenance mode on all pgdog pods, do a live config reload on each of them, then disable maintenance mode (this is to make the change atomic between all the pgdog instances). Downtime there is about 2-3 seconds and all it does is make new SQL requests from existing clients wait, it doesn't break the connection or anything.

The sloppa in this article is... offending.

The obvious AI headings, pointless genned image of people (I'm starting to think islam had a point with discouraging depictions of human figures), and especially the blurry, artifacted, distractingly skeuomorphic diagram, with random wire traces going everywhere... this is a technical blog, not an investor sales pitch! Every time I see one of these, I have to double-check for a second if I'm not on some phishing SEO site!

If even Google, previously a gold standard of technical writing, is falling prey to this kind of laziness, then I have nothing to worry about -- knowing how to write without a language model in the driver's seat is gonna be a top tier skill in the future...

A damn shame too, as I've been following the progress of JXL in the standardization pipeline for a few years now and was quite interested in the historical breakdown, but all that's gonna stick with me from this is the disrespect I felt as a reader.

... with a high-resolution scan of the work itself available for download, to boot. I really appreciate whenever museums go out of their way to share those publically! Much better than many paintings only officially available as some 400px thumbnail.

Oh hey, Tixl on HN! It's probably my favorite piece of FOSS creative software, and honestly blows After Effects out of the water for me (as a hobbyist who likes cooking up some vis for my DJ sets).

I'm not kidding when I say it has the potential to become the blender for 2D/3D VFX -- the node engine is super powerful, the primitive building blocks are all well thought-out and integrate with each other very nicely, the performance characteristics are amazing (all optimized for realtime!), and there's a ton of I/O for everything from mouse input to OSC/MIDI, camera control, elaborate audio reaction... and also just plain TCP/UDP/HTTP/Websockets! It's such a powerful glue piece, but also tons of fun to mess around with on its own.

The best part? You can create your own components, define your inputs/outputs, and compose them together. The even bestest part? You can dig into the predefined components/effects and see how they work, as they're very often implemented in the same way! The visual editor all drills down to C# in the end, and you can drop into the code or write some HLSL shaders if you want, all with hot reloading.

Just give it a try, you won't regret it :)

How relevant is this (and the NSA's general spying capability) in 2025?

We hear a lot about local agencies perusing the services of private companies to collect citizens' data in the US, whether that's traffic information, IoT recordings, buying information from FAANG, etc. What's the NSA's position in the current administration? (e.g. we've heard a lot of noise in the past about the FBI and CIA getting the cold shoulder internally. I wonder how this applies to the NSA.)

Palantir is firmly cemented into military-industrial infrastructure, and business is booming, but Karp is not letting up. He has said he wants Palantir to be as dominant and indispensable as IBM was in the 1960s, when it was the world’s largest computing company and shaped the way government and private companies did business.

I was thinking even before this line that he gives off the impression of really admiring IBM, especially their German stint from 1933 to 1945.

I finally took the plunge and upgraded from Ubuntu 22.04 to 24.04 on my work laptop. With that come all the gnome package updates and such.

I reboot, load up my session, and a little while after need to grab a couple files from a zip archive. I double-click it in Nautilus, nothing happens. I give it a couple more clicks before I suspect something broke, right click it, and see "Extract" as... the default cursor action...?

I go back up and see five fresh copies of the folder that was inside the zip. I delete them all, go back to the file itself, right click, open with > file-roller. I try and drag'n'drop the couple files I need: doesn't work, for some reason. Great, they've broken drag and drop, too.

I look it up, stumble on https://gitlab.gnome.org/GNOME/file-roller/-/issues/4, 7 years old issue -- I can already tell this is gonna be a joy; scroll down some, see https://gitlab.gnome.org/GNOME/file-roller/-/issues/4#note_1..., and audibly groan.

is drag and drop extraction from Nautilus (or other file managers) really that important?

Why, yes it is! It's even worse if you go and _break_ drag'n'drop support on X11! I'm not even using Wayland!

But, oh well, looks like file-roller is unmaintained and outside of core Gnome scope now. Nautilus' zip capabilities are enough, they say! Why would a user want to inspect the contents of a zip archive before wanting to extract it, or god forbid select specific files, after all? Definitely not worth keeping as a core OS/DE feature.

And the PR on file-roller that fixes this on wayland with... a custom fuse virtual filesystem?! has been untouched for the past 2 years, never to be merged.

I'm moving to KDE, thank you very much.

The OOM issue is fixed in the most recent updates -- it was a memory leak linked to the file scanner, IIRC.

Sakugabooru's blog has much information on the internals of anime production. This article answers some questions about the production committee model: https://blog.sakugabooru.com/2017/05/02/what-is-an-animes-pr...

Otaku/doujin culture, and the creative industry that rose around it in Japan, is as good as it gets when it comes to finding good ideas and propping them up. Basically anyone who can draw can release their own manga/webnovel/illustrations on pixiv, twitter, and others, get a couple volumes out with this or that publishing house, see where it goes and whether the public catches on. Self-publishing plays a huge part in this, whether it's doujinshi (self-made [often derivative] books sold at conventions like Comiket, Comitia, or online, with a substantial proportion of r-18 [but not only] content), doujin music albums, indie games and visual novels, etc.

Funding for anime adaptations is plentiful, and fan support helps bridge the confidence gap where production committees (consortiums of multiple publishing/IP companies pooling the money to distribute the risk) won't go for more indie / experimental works. Profit is recouped on developing the brand and merch, while leaving plenty of room for directors and studios to establish their own auteur identity. Studios are getting leaner and more focused these days, splitting off into smaller entities kicked off with a project or two with more margin for talent to shine.

It's not all perfect, though. Freelance work and lack of mentoring has really put a dent into the supply of new animators, who lack job security and often swing between studios left and right, but there are industry efforts to fix this and preserve knowledge, with the oldest pioneer animators now starting to hit their 60's and 70's.

Question to anyone with experience in this domain: I have CSAM spam problems on a forum I host, with bots putting link shortener URLs embedded in images rather than the post body. Traditional OCR software deals poorly with them due to font modifications and intentional text edge modifications, and I'm obviously not gonna use a SaaS/closed source model to upload a bunch of may-be-may-not-be-CSAM pictures, so looking for a way to do this locally, with cheapish inference if possible (I don't mind spending a minute of compute to get the result out for one image, but need to do it on the CPU).

Is there any small model that would do this effectively, with pure text extraction (without going for any kind of formatting or whatnot)?

Mixxx can stream to an icecast output or similar, you'll need to setup the stream server separately though.

Been using it for the past few years, nothing bad to say about it, lovely piece of software. Vendor lock-in is very present in this field, with different brands of controllers supported by a myriad of proprietary DJ applications all more interested in onboarding you to their music subscription services rather than implement useful features or support open protocols.

Meanwhile, Mixxx allows you to write your own adapter scripts for any controller you have (as long as it outputs MIDI), and there's a built-in library featuring scripts for the most common commercial controllers and MIDI devices out there.

Might be wise to make it more apparent that you need to close the right sidebar before you can do anything on-screen -- I tried clicking on the "Get Started" buttons, found them to be unresponsive / non-interactive, tried clicking on stuff on the side, and closed the page at first, thinking it wasn't tested on Firefox.

This Japanese stereotyping would be a little more credible if Sony Entertainment's headquarters hadn't moved to California years ago...

I'd like to get a VDI infrastructure setup for me and my partner -- something that enables both of us to run our computers with a single big machine, and not have cables running everywhere through the house, while being able to sit wherever we want and use our infra from any given place.

At the moment, I've brainstormed:

- A main server that runs some type 1 hypervisor (Xen or Proxmox, will need to see which is more adequate)

- Light "client" devices (laptop, for example), that may either be connected in a wired manner to the server (e.g. separate desks), or remoting into it through Wireguard. Each desk will feature a KVM-style setup with a docking station that offers screens, keyboard, and a range of USB ports.

- Individual VMs for running our respective OSes to our preference, some flavor of linux distro. Inputs from the client device (e.g. USB, Keyboard, Screen) should be forwarded / matched to the VM.

- A windows VM for gaming, running two sessions for each of us: GPU passthrough is a must. I would like to make use of Looking Glass somehow, if possible either through the Linux VM on the same server, or on the client machine. The latter would probably be better for performance, I suppose, given you don't have to forward input devices twice... but I'm also worried about whether the buffer-copy mechanisms from Looking Glass would work with such a setup.

So far, I'm looking into Moonlight/Sunshine as a general desktop redirection setup: my hope is that I can pass something close to direct framebuffers on an ethernet connection while at home, and switch to compression while I'm away, hoping to achieve as little latency as possible in all cases (so giving absolute priority on the host to the streaming process, if possible, kind of like an RT system). One notable thing is that Sunshine by itself doesn't support generic USB redirection. Has anyone tried using usbredir for this purpose?

In general, it's hard to find relevant information for this kind of home hypervisor setup with a focus on gaming/latency and general transparency all around... would appreciate tips if anyone's attempted something similar before. Thanks!

are there codecs that have much better (perceptible) quality than Apple AAC 256kbps (or achieving similar quality at, say, 160kbps?)

Opus achieves ABX transparency at around 128kbps (as in, the threshold where the vast majority of users taking a fidelity test are unable to tell the difference between the opus-encoded and lossless version).

https://abx.digitalfeed.net/opus.html

Considering the precedent in circumventing decryption issues by calling out to a library (https://en.wikipedia.org/wiki/Libdvdcss), what is the likelihood that someone just builds a libxcidecrypt and development of yuzu continues unimpeded on a fork (somewhere like Europe for example where the DMCA doesn't hold)?

Similar thing in France, however related to degrees: a Master's Degree fits within the common framework of European tertiary studies, AKA your 2-year graduate cycle that ends up totalling 5 years if you include a standard undergrad cycle (Licence). However, a "mastère spécialisé" (which is an actual trademark) is not an officially recognized diploma at all, but a label used by schools to denote (usually paid and expensive) "specialization paths". They provide no academic credits, and you can't move forward to a PhD with one. Many prestigious schools (grandes écoles) provide those kind of certifications to clueless foreign students who end up paying huge sums of money for something that holds no formal weight in any academic framework.

There are some useful cases — for example, if you're taking a rather bloated image as a base and trimming it down with `rm` commands, those will be saved as differential layers, which will not reduce the size of the final image in the slightest. Only merging will actually "register" these deletions.

fsearch is the best locate front-end for Linux, but sadly, I've got many gripes with it... crappy drag'n'drop, no daemonization to minimize to system tray, closing the app resets the clipboard for some reason (EXTREMELY annoying when you open it to copy a file path), the list is long. Not to mention that locate itself doesn't auto-update the index with fs changes.

I miss Everything :(

As other people have said, it's a shame that so much of psytrance relegates itself to being "electronic dance music", playing off the same clichés and subdividing itself so much such that the most minute production details are catalogued. While you could reasonably say that about any electronic genre, Psytrance really ups the level of having any given median track sounding indistinguishable. Yeah, the sound design experimentations are cool; yeah, it's a good sound... but I wish I would actually remember it for _something_ specific it does well.

In general, I think I miss the songwriting aspect most. Infected Mushroom really nailed it, which is why they're so popular. Nowadays, I can only find the Japanese doujin music scene among those that regularly mix and match psytrance with various other genres.

A few other recs going more in that direction:

- Talpa (https://sundancerecords.bandcamp.com/album/when-the-somberne...)

- Jerico's Law (https://juno909.bandcamp.com/album/another-law)

- Migma Shelter (https://www.youtube.com/playlist?list=OLAK5uy_l3dfWAOgloDDde... -- j-pop psytrance!)

- Hilight Tribe (https://hilighttribe.bandcamp.com/album/temple-of-light)

- Koxbox (https://koxboxmemo604.bandcamp.com/album/the-great-unknown)

- Hallucinogen (https://hallucinogenmusic.bandcamp.com/album/the-lone-derang...)

- And finally, the best hi-tech work I've ever heard. Sephax - Electric Currents (https://alice-drecords.bandcamp.com/album/sephax-electric-cu...)

I personally use a file provider for the dynamic traefik configs (yaml files) loaded from a bind mount in the same folder I keep my compose. Auto-reload on changes, and it makes it clear what I'm routing to and from by having proper indentation for my router, service, and middleware fields. And since everything is in the same network, I can just define the container name as hostname -- the DNS entries are automatically created.

A Czech linguist of Slovak descent, Karol Moravčík, identified him as Hungarian, and on 11 August 2000, Toma arrived back in Hungary where his family was identified through DNA matching.

Someone else recognized the language he was speaking, and that it wasn't actually gibberish (and so he wasn't some crazy old man).

Please don't reencode your entire collection to "save space", or use git for version control.

Put your lossless files on a server as a source-of-truth (with a regular cold backup somewhere else) and install a streaming server, like Navidrome[1], which will allow you to transcode on-the-fly to all of your devices. This is how you build a true "digital music collection": so that you won't regret it 5 years from now, when the site you bought your flac's from closes down/erases your files (as has happened before on Bandcamp when an artist account gets deleted!), leaving you solely with the reencoded mp3/opus files you kept, unable to move to better formats as they progress.

[1] https://www.navidrome.org/

Careful with performance if you're on an underpowered machine, as with all content injectors, it could provoke frequent freezes if you have a decent to large amount of tabs open.

https://git.synz.io/Synzvato/decentraleyes/-/issues/323

So what's the link to undivided attention?