Hacker Newsnew | past | comments | ask | show | jobs | submit | jszymborski's commentslogin

There are some technical barriers to approaching fediverse platforms, but I personally see the main barriers being cultural.

I'm a big proponent of Mastodon and still love using it, but the culture (especially early on) was exceptionally protectionist and lots of people got bullied off for very silly reasons. I think the attitude is less like a children's secret club and more chill generally.

All this to say, I think this is will get better, but the best way to help the fediverse is to join it, be active, and be chill.


I also don't _love_ vibe coding and do it just for exploration/recreation, but I also have long thought an LLM trained and tuned specifically for a language that is best for LLMs might be ideal.

Currently, using Claude to vibe code Rust is _much_ more hit-or-miss than using it for Python... so Python has become the lingua franca or IR I use with it.

Often I'll ask Claude to implement something in Python, validate and correct the implementation, and in a separate session ask it to translate it from Python to Rust (with my requirements). It often helps.

Claude is particularly bad at hallucinating the APIs of Crates, something it does a lot less for python.


It's also already used for language modelling:

MLM is masked language modelling, another phrase for training models on the cloze task. It's the most common way to train encoder-only models.

CLM (causal language modelling) is the other common task where you autoregressively predict the next token given the previous ones. It's the most common way to train decoder-only models.


> ...they're family owned

Well that's your problem there.

I do overall agree that Valve is only situationally the good guy here, but they do also have a sustainable approach to business and growth which I think helps this.


They said this about gaming too... all it takes is Valve-esque sponsor

Until Valve starts shipping native SteamOS games, they were and are right.

Games are developed on Windows, using Windows APIs and development tools, and then Valve does the job studios don't see any value in spending development resources on, even though some of them use engines that also target GNU/Linux.

Talking about Linux gaming with Proton is no different than if Windows users would start talking about Windows being their favourite Linux distribution due to WSL 2.0.


It takes a stable ABI and a single target for apps. Don’t expect music production to ever show up for gnu/linux. Maybe Android

Bitwig is available for Linux, but I don't know what the limitations are.

A lot of VSTS aren't going to work even with tricks like WINE.

They often have complicated auth systems, etc.

I don't really want to learn a new DAW.


> They often have complicated auth systems, etc.

The paranoia audio devs have with piracy is insane. Ironically I think the worse your DRM is, the more your product should be pirated :)


It's often 20% DRM and 80% marketing.

Ohh cool, this plugin is "free". Looks like I have to register for an account and install a bunch of unrelated crap I really don't want.

The up selling never stops.

Then again. 30 years ago a lot of these trucks were straight up impossible without spending tens of thousands.


A lot of people give high praise to Yabridge, but I myself haven't tried using it for my existing VST library and just found alternative plugins that work with linux.

The state of things isn't great IMHO. Im not sure I trust any of EncFS, CryFS, and gocryptfs.

Many leak metadata and/or have serious security concerns.


Metadata leakage is a fundamental issue when you go from block to object. I can think of some schemes that would help but they’re all kinda nasty lol


Of course, and I didnt intend to downplay the efforts of those projects. Just pointing out that they don't meet the requirements of most threat models.


I kinda gotta push back on that

Most threat models don’t include state level or equally well funded/motivated actors.

Some of those, in theory, are fine for most corporate usage - when used or implemented by knowledgeable people. Shipping it as a consumer product is a bit rougher of a story, although most companies seem to cope by not giving a shit (lol, oof)


Can you detail the current metadata and security problems with CryFS? Do they also extend/apply to securefs?


I was under the impression that thr US Military doesn't use GPS; that it was a civilian thing and that they had something more precise and secure.


The US Military does use GPS, but ordinary civilians don't have access to high accuracy data. Commercial vendors can license for higher accuracy, but its a hybrid civilian/military system with higher quality data for military use cases.


I think that was previously the case, but not anymore. I can get within a couple meters accuracy on my phone, which is plenty good enough for weapons guidance.


Not since May of 2000.


GPS was built by the US military. Its existence was first revealed in Desert Storm back in 1990 when the US military used it to drive tanks into the desert, then come out where and when they wanted. The Iraqis were quite surprised, since they knew that desert, and knew that they would get lost if they tried to do the same maneuver.

Parts of the GPS signal are encrypted to be only useful for the military. The result is that civilian systems an average 4.9 meters of accuracy, while the military is precise to something like a meter instead. But that extra accuracy doesn't help if the signal is jammed.


It was revealed earlier than 1990. In 1983 it was publicly announced that it would be made available for civilian use. Hard to do that without its existence being revealed.


GPS was always intended for both civilian and military purposes, dating back to its inception in the 1970s. In response to the downing of KAL 007 by the Russians in 1983, civilian access to satellite navigation became an explicit guarantee. But that wasn't really an operational turning point, just a clarification of policy.

Before KAL 007, GPS was unaffordable to most users, and not widely developed commercially because there was still uncertainty about whether it would remain available in the long run.


Huh.

I remember the capability coming as a surprise in 1990, but you're right that Reagan announced that civilian access would happen back in 1983.

Now I don't know if I was misremembering, or if Iraq was simply unaware of the technology, or whether that announcement was for access at some future date.

I do remember discussing GPS on sci.physics in the mid-90s though. Where I learned that GPS is the only commercial technology that has to take general relativity into account. Clocks on Earth run measurably slower than clocks at the altitude of GPS satellites, and the effect is big enough that GPS has to correct for it.


I wonder if you mixed it up with the F-117. It was technically public before the war, but got a lot of attention during it.


No, very much not. It was an article about how shocked the Iraqis were that the US tanks were able to disappear into the desert, organize, and then assault where and when they wanted to.

Looking back at the history, the tanks began to go into the Saudi desert in August of 1990. They then launched their massive assault on Feb 24, 1991. And caught the Iraqis completely flatfooted. With the tanks moving faster than the news of the tanks for several hours.


They use a different coding scheme on the same physical signal. It is resistant to spoofing but far from immune to jamming.


It was developed by and for the military and is still managed by the military. Those are not normal qualities of civilian only systems.


Codeberg or GitHub pages are free. For static website hosting, NearlyFreeSpeech.NET is... well... nearly free.

https://www.nearlyfreespeech.net/services/pricing


Netlify is also a good free option for non techies since you can just drag a directory to deploy it. I’ve also used Cloudfare Pages.


Likewise for GitLab, and a "nice" url <username>.gitlab.io



HTTPS ain’t cheap though.


What do you mean? I don't think HTTPS is a paying feature of sdf, and HTTPS is otherwise free thanks to let's encrypt.


MetaARPA tier membership (quarterly fee) is required to have HTTPS on your personal website - personal sites hosted on the main BSD cluster don’t have it.


I alsi feel this way about point n' click adventure games. Something about pre-rendered graphics that makes me feel great.


The posted article which you are commenting on is entirely about why you shouldn't...


It doesn't address the comment to which I was replying.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: