22c

On the idea of replacing ones self with a shell script, I think there's nothing stopping people (and it should probably be encouraged) with replacing ones use of an LLM with an LLM generated "shell script".

Using an LLM to count how many Rs are in the word strawberry is silly. Using it to write a script to reliably determine how many <LETTER> are in <WORD> is not so silly.

The same goes for many repeated task you'd have an LLM naively perform.

I think that is essentially what the article is getting at, but it's got very little to do with MCP. Perhaps the author has more familiarity with "slop" MCP tools than I do.

There is at least some awareness already when it comes to the performance of the regex engine:

https://github.com/openai/tiktoken/blob/main/src/lib.rs#L95-...

I read the article, I summarised the extremely lengthy points by using AI and then replied to that for the benefit of context.

The HN submission has been editorialised since it was submitted, originally said "Yes, I will judge you for using AI..." and a lot of the replies early on were dismissive based on the title alone.

I think it's a good idea, it does disrupt some of the traditional workflows though.

If you use AI as tab-complete but it's what you would've done anyway, should you flag it? I don't know, plenty to think about when it comes to what the right amount of disclosure is.

I certainly wish that with our company, people could flag (particularly) large commits as coming from a tool rather than a person, but I guess the idea is that the person is still responsible for whatever the tool generates.

The problem is that it's incredibly enticing for over-worked engineers to have AI do large (ie. diffs) but boring tasks that they'd typically get very little recognition for (eg. ESLint migrations).

Hmm kinda makes sense to keep them separate because the agents perform differently, right?

You might want to tell Claude not to write so many comments but you might want to tell Gemini not to reach for Kotlin so much, or something.

A unified approach might be nice, but using the same prompt for all of the LLM "coding tools" is probably not going to be as nice as having prompts tailored for each specific tool.

The entire article, probably quite intentionally, seems to overuse semicolons (in my opinion). I say this as a semicolon enjoyer, but I think the overuse of semicolons in this article leads the reader to a bit of semicolon fatigue by the end of it.

Someone better versed with parsing crab-grade enterprise-ready Rust could probably give more insight but I think it boils down to a lot of

    assert 1 != 0;

kind of lines...

Part-way through the section on bit-twiddling, I thought to myself "Oh I wonder if we could use a solver here". Lo and behold, I was pleasantly surprised to see the author then take that exact approach. Love the attention to detail in this post!

French door has less insulation because there's more "seams". Also typical french doors tend to have an ice machine or water dispenser, which is even less insulation and/or more seams.

Not sure how it works in the US but many countries can choose to deny entry for any reason and my understanding is they don't even need to provide a reason.

If that is the case here then I really doubt the researcher was told that they have been denied entry due to their personal opinions on the Trump administration. This really just leaves us to speculate.

Nice post and clearly some talented engineers, props for giving things a try.

Interested to see what you two cook up next!

Sadly, a modern day french door is pretty inefficient. Consumers prefer them because of their looks but they are very far from being an efficient way to keep food cold.

It's not immediately clear from the documentation but if anyone is familiar with Iroh, I'd love to know if it's suitable for use in disconnected networks, eg. WiFi networks or PANs with no internet.

I suspect that Iroh is primarily designed for P2P between two devices on the internet.

I don't disagree but I think some of these problems could potentially be solved by having somewhat of a birds nest of a filesystem for large blobs, eg.

/blobs/<sha256_sum>/filename.zip

and then symlinking/reflinking filename.zip to wherever it needs to be in the source tree...

It's more portable than hardlinks, solves your "source of truth" problem and has pretty wide platform support.

Platforms that don't support symlinks/reflinks could copy the files to where they need to be then delete the blob store at the end and be no worse off than they are now.

Anyway, I'm just a netizen making a drive-by comment.

or, you know.. symlinks

I wonder if Zedeus would be willing to list some reasons why they have decided to ditch Nim. 6 years is a long time using a language, I'm sure there's some useful feedback.

256k context window and FIM support sounds good. I can't see it mentioned on the release page how big the model is, they say sub-100B but they are comparing it to 22B and the 22B seems to hold up quite well against it. If we're talking 80B vs 22B then it seems like quite a heavy-weight model.

As a polyglot, sometimes the existence of a certain library or interface greatly influences what language I decide to write something in.

Twig is indeed time-tested, but I don't see a good reason to use Twig unless you are already using Perl. I was kinda hoping this would give reasons to use Twig over proven XML libraries from other languages, but I suppose it makes sense given this article was published to a Perl specific website.

Nice write-up, though!

> Once you add more than one character per row its all over anyways.

Ah, but that can be solved by returning the carriage to the start of the line :)

Thus giving you the perfectly valid control sequence: LFCR!

Seems to be a lot of hype around this project in certain niches (if "niche hype" is something that makes sense), but I've only just realised that the same devs were involved in the Norns project so it has ticked my interest up a notch.

The UI and form factor look really nice, but the asking price is currently the biggest thing on my mind.

I wish the project success, regardless!

Interesting how the narrative being pushed by companies spending a lot of money on AI always seems to be gravitate to around 20% of time saved. I've not seen anywhere how this figure of 20% is determined.

It seems to be a number that the masses are willing to just accept without any reasoning.

If they said 30% people might start to ask how, if they said 5% people might just shrug and think it's not significant.

I think the proposal is fundamentally flawed because of this reason.

Facebook, Instagram and X/Twitter are probably what's intended here, but what about Tumblr, DeviantArt or Discord? What about Reddit or a generic forum? What about VRChat or Webfishing?

If this is about protecting children from harmful depictions of body image or misogynistic content, then why not instead propose a law that states online services that allow children to join need to appropriately moderate the content that is shown to children or could face massive fines. I don't necessarily agree with that approach, but at least it would make sense with what their stated objectives are.

I like the idea, but it seems to have come a little too late to save everything on Flickr.

Flickr themselves started purging any public images over a certain amount on free accounts.

I only found out after seeing a Flickr email in my spam inbox after several months of the start of the purge, so some of my images had already been purged.

I suppose their approach was still preferable to deleting everything at once.

The soundtrack for The Neverhood was also very nice.

That is the one, thank you!

I remember reading a blog post many years ago that stuck with me, I don't remember the title of the post or who wrote it but it was essentially along the lines of "beware when companies start to take away the free snacks". It made a pretty convincing argument about how even seemingly minor changes like that start to signal a greater shift in the companies attitude toward how they want to treat and be perceived by their staff.

I think a lot of Amazonians, even those who choose to stay, will look back on this as their "no more free snacks" moment (I also think there are probably many Amazonians who are already past that point, but that is a different topic).

I don't think Amazon is making a good decision here, but what makes you think the existing workers who are happy to work from the office would not be willing to work from home during a crisis if asked?

A global crisis necessitating a transition to working from home should not immediately result in a company like Amazon needing to substantially hire more staff, in fact, that is probably a big part of what got them into this situation.

I think you've hinted at it, but years ago I was interested in learning how to draw and was given the advice of drawing poses on a timer (I think it was 120s but the goal is to eventually get it down to 30s or less).

There were websites back then which would automatically flip to the next pose once the timer ran out, you'd have to start a new page and sketch it out again.

If you do this for even 10-15 minutes a day every day, you eventually start to see an improvement on your ability to draw poses (starting with picking up on which lines are the most important to convey a silhouette, eventually moving into things like muscles, bone structure, etc.). You can even refer back to your earlier drawings to see the progression.

I never got any good because I didn't stick with it, but I did get a lot better than how I was when I started.

Related discussion: https://news.ycombinator.com/item?id=41791369

I think this is a reasonable response if it is made in response to what the Tweet author linked.

Users are confused as to whether or not paying for a license for ACF Pro (a highly popular custom forms plugin for Wordpress, published by WP Engine) constitutes as being "economically related to WP Engine" and might impact their agreements with Wordpress.org

In response to this confusion, Matt Mullenweg told those users to consult an attorney. Unfortunately for most users, consulting an attorney before agreeing to login to a website is not a financially viable option.