You’ve insufficiently described any technical reason to use git as the basis for your chat history. Git is a poor choice for all the reasons I explained in your last post and more. It’s like using paint as glue or using a microwave to heat your bathwater.

While it might be technically possible and work on first glance, there are few to no technical advantages to doing so, and anyone could tell you as such. In fact in some cases it might even be a detriment.

In your posts, you’re talking about the usage of Tor and how WebRTC is a bad fit compared to Git and I can only guess at how you reached the conclusions you did.

Literally just ask your AI why using Git as the basis for a secure encrypted messaging application is a bad idea and I’m sure it could do so more eloquent than I.

Like, my first thought is “why not use https or even TCP directly rather than Git” sqlite and a TCP socket is more sane than this.

I’m actually not against vibe coding, but this is the kind of stupid shit people point to when they say vibe coding is bad.

Git is designed as a permanent, historical record of text documents. You don’t “clear” it out by deleting stuff, committing, and pushing. That stuff still exists, even if you change commit history it’s going to stick around in the reflog for a long time.

GitHub has a maximum of something like 10GB of storage available. That being said, I think your described requirements don’t actually match git. If you’re just publishing messages publicly (even if encrypted) then you can do the same thing with a simple http server.

I think it’s a really poorly thought out idea.

I don’t think this is ai written. It seems like a very stream-of- consciousness diary rather than an attempt at making anything understandable.

I think I take issue with the idea that it’s the “chat” interface that makes people shut off their brain and stay in that groove. I feel that the answer is more the very nature of LLMs being one of our attempts to get AI to pass the turing test. Continuing the conversation is such a core part of what an LLM is beyond the UI. If it made you think, you wouldn’t believe it to be a human.

We use Ruby at work, for a team I’m only tangentially involved with, and I’m not a fan. I’d be curious to know what a company that uses it as their primary language looks like. It seems like there’s a lot of funky edge cases especially in templating that have bitten us.

Has it really been 10-15 years since Ruby peaked in popularity? I never really paid much attention but I feel like time is a fiction.

I’m not appealing to authority. I’m pointing out that they’re not a bot. As I said in my first post, I wouldn’t claim it’s entirely human generated, I have no idea one way or another. What I said to start with and now is that it’s worth addressing the Blogpost on its merits not on its use of emdashes.

It’s crazy to me that you think I’m even saying there smart/clever/whatever and could do no wrong. The only thing I’m appealing to is the evidence that they exist.

The author has been a CPPCon presenter, worked at Bloomberg, and has a GitHub history going back at least to 2022 where they were contributing to LLVM. This isn’t coming from their website but from the actual websites.

Based on their history and experience that’s public on their profiles from an era pre ChatGPT, I think you are overreacting and making a fool of yourself.

It’s pretty scary how we live in an era where semi competent writing gets assumed to be AI and you refuse to fact check yourself before shitting on them as being not real.

Honestly it’s a pretty well written and explained article. I hesitate to say that it’s entirely human generated, but I certainly would try and judge it for its content before just writing it off as slop.

Personally I think the author overstates the practicality of using counters for serialization, as the ordering of source code is also a very cosmetic thing and tweaking scenes ruins the save format. There’s no possibility for backward compatibility either in such cases.

It’s hard for me to imagine which direction it would end up going when right now the quality of AI output is so varied.

I think there’s also something to be said to just “doing things” that are in the public eye.

What I mean by that is that (part of) the reason I got the job I’m at right now is because I happened to do something at an org that my boss worked at years prior. It wasn’t a “connection” in the sense that we knew each other or ever crossed paths, but it still got me a foot in the door.

If you can make a meaningful contribution to a well known project or organization that will open doors via “connections” that you never knew existed.

I think having “LTS” mean nothing is counterproductive. Why would anyone agree to apply “meaningless” labels like that? It would only confuse people more if we applied a label which implied support but nobody was actually obligated to support it any more than another release.

It’s a bit of a mixed bag for me to be honest. I understand the desire to accurately package dependencies and maintain control over the dependencies without relying on a third party host, but I feel like everyone (whether that’s rust, node, python, ruby, etc) should just maintain a separate registry. As in not package it in the “Debian repos” but a debian mirror of crates.io that the debian maintainers maintain. To whit I can just download the Debian rust toolchain and have it be pointing at the Debian rust package mirror.

I’m sure there’s a lot of extra infra required for something like that, but I genuinely believe it’s more sane to try and get these languages to adopt that, rather than what we have now.

I could be easily swayed another way, but that’s how I feel currently.

Phoronix comment sections never fail to be cesspools

The stress-ng package is like 5MB.

There are a lot of possible ways to validate this, namely just downloading a benchmarking tool and using it. Is there a reason you don’t want to do that?

From a completely unscientific but ‘experienced’ perspective I think the problem is that life just gets in the way as you get older, and you prioritize your own life rather than trying to learn.

Whether neuroplasticity means you can learn things later or not, the opportunity to learn things later just isn’t there without effort.

Having a job, kids, a mortgage and no social obligation to learn in a structured and organized way probably impacts you more than anything neurological.

I know, it’s pretty awesome right? It’s one of the highlights of the community in my opinion.

I read the first book or two series in middle school(‽) and it was pretty good, if I recall correctly this had a pretty good sibling relationship, and the only thing stopping me from finishing it was the library didn’t have the rest.

Definitely don’t just dismiss books for YA though! There’s no reason you can’t enjoy a good book because of the target audience. I still re-read my favorites all the time, and try out new books.