davely

> you’re running arbitrary third party code on client machines without any form of sandboxing. This is a level of security you would not deploy into Azure, so why is “run arbitrary 3p code on someone else’s machine” appropriate for VSCode?

More and more, I am starting to think I need to run my development environment (for both work and personal projects) in a VM.

I am on MacOS, so UTM or Parallels would work pretty well I think. Sadly, I think my work explicitly forbids us from running VMs or accessing our services from them.

I'm in the middle of a particularly nasty refactor of some legacy React component code (hasn't been touched in 6 years, old class based pattern, tons of methods, why, oh, why did we do XYZ) at work and have been using Aider for the last few days and have been hitting a wall. I've been digging through Aider's source code on Github to pull out prompts and try to write my own little helper script.

So, perfect timing on this release for me! I decided to install Claude Code and it is making short work of this. I love the interface. I love the personality ("Ruminating", "Schlepping", etc).

Just an all around fantastic job!

(This makes me especially bummed that I really messed up my OA awhile back for you guys. I'll try again in a few months!)

Keep on doing great work. Thank you!

I often read these threads on HN and they make me feel like that I am perhaps a horrible software engineer or just weird…

I _like_ building things with React.

I think we need to wait for someone to convert it into a GGUF file format.

However, once that happens, you can run it (and any GGUF model) from Hugging Face![0]

[0] https://huggingface.co/docs/hub/en/ollama

I recently took their CodeSignal assessment, which is part of their initial screening process.

Oh, wow. I really believe they are missing out on great engineers due to the nature of it.

90 minutes to implement a series of increasingly difficult specs and pass all the unit tests.

There is zero consideration for quality of code. My email from the recruiter said (verbatim), “the CodeSignal screen is intended to test your ability to quickly and correctly write and refactor working code. You will not be evaluated on code quality.”

It was my first time ever taking a CodeSignal assessment and there was really no way to prepare for it ahead of time.

Apparently, I can apply again in 6 months.

> Interviews aren’t about solving problems.

Eh, I wish more people felt that way, I have failed so many interviews because I haven't solved the coding problem in time.

The feedback has always been something along the lines of "great at communicating your thoughts, discussing trade-offs, having a good back and forth" but "yeah, ultimately really wanted to see if you could pass all the unit tests."

Even in interview panels I've personally been a part of, one of the things we evaluate (heavily) is whether the candidate solved the problem.

> Wanting to be paid under false pretenses is the definition of fraud.

What? No, it isn't.

Regardless, if the job requirements state "X years of XYZ experience" and you have to have >X years of experience, then using AI to look up how to do a leetcode problem for some algorithm you haven't used since your university days is absolutely not "false pretenses" nor fraud.

It will take a long time, but I think the ever growing usage of GenAI tools will ultimately make people desire human created works even more.

That said, there's a common thread in a lot of writing similar to this piece. Basically, it's some form of, "GenAI isn't art. GenAI creations lack a human element present in the creation process. GenAI creations cheapen the creative process in general due to the ease at which they can pop out new things."

I'm mixed on this. This is something we've always had to deal with as a species. This greatly simplies a lot of human history here, but consider:

- Our (long, long ago) ancestors had to paint on walls with their fingers. They had to make music by banging sticks together

- Paint brushes were invented. Musical instruments were invented. These things make it easier and quicker for people to _create_

- Mechanical machines made it easier to make garments and print books

- Digital machines made it even easier to write stories, or make music, or play music

The word processor alone made it ridiculously easy for highly evolved apes to start pounding on keyboards and outputting works of various lengths and quality.

Prior to the GenAI craze, there were music creation tools that contained multi-gigabyte sample packs. You could throw together random samples and make good sounding songs.

Photoshop and Illustrator made it exceedingly easy for artists to flex their creative muscles and get their ideas out there.

At the end of the day, these are all tools. And I think GenAI will ultimately be used in a similar way. It's a tool. Can it be used in a bad way? Sure. Just like PhotoShop. Just like Word.

Regarding what is and isn't art -- I don't think it's up to us to decide. And does it matter? History is full of ridiculous examples of things that might or might not be art.[0][1] Humans don't necessarily need to be involved in the creation process either.[2]

[0] https://en.wikipedia.org/wiki/Fountain_(Duchamp)

[1] https://en.wikipedia.org/wiki/Comedian_(artwork)

[2] https://en.wikipedia.org/wiki/Pierre_Brassau

For what it's worth, HuggingFace provides documentation on how you can run any GGUF model inside Ollama[0]. You're not locked into their closed library or have to wait for them to add new models.

Granted, they could be a lot more helpful in providing information on how you do this. But this feature exists, at least.

[0] https://huggingface.co/docs/hub/en/ollama

I think it's important to bring up the face that llama.cpp has an MIT license[0]. Notably, the MIT license "permits reuse within proprietary software, provided that all copies of the software or its substantial portions include a copy of the terms of the MIT License and also a copyright notice.[1]"

You'll find that Ollama is also distributed under an MIT license[2]. It's fine to disagree with their priorities and lack of transparency. But trying to argue how they use code from other repositories that permit such a thing is tilting at windmills, IMHO.

[0] https://github.com/ggerganov/llama.cpp/blob/master/LICENSE

[1] https://en.wikipedia.org/wiki/MIT_License

[2] https://github.com/ollama/ollama/blob/main/LICENSE

You should re-read his tweets from January 6th[0]. He didn’t tweet about peace until hours after things started, (he was gleefully watching it unfold on TV[1], remember), and when he did, it was a half hearted one line sentence to go home.

And three of his tweets were deleted by Twitter, including this gem from late in the day:

“Mike Pence didn't have the courage to do what should have been done to protect our Country and our Constitution, giving States a chance to certify a corrected set of facts, not the fraudulent or inaccurate ones which they were asked to previously certify. USA demands the truth!”

This doesn’t read like someone who was remorseful about the actions that happened that day on his watch.

0: https://www.presidency.ucsb.edu/documents/tweets-january-6-2...

1: https://www.reuters.com/world/us/us-capitol-probes-season-fi...

Oh, yeah. I asked QwQ (their text based reasoning model released a few weeks ago) about the Tiananmen Square Massacre and got a simple response [1]: “I’m sorry, I can’t help you with that.”

[1] https://bsky.app/profile/davely.bsky.social/post/3lc6mpnjks5...

I’m not sure how that refutes my point. Of course they are.

That probably explains why the Amazon website just took 5 seconds to load on my phone while on a gigabit connection at home.

Every single React app has to ability to be nice and performant.

The issue (and this isn’t React’s fault) is that every single web app is so obsessed with measuring analytics and engagement metrics and loading multiple ad containers and and and…

Whether Vite, or Svelte, or Vanilla JS, when you start stuffing a site with all these tracking pixels and other junk, they are all horrible.

There is likely a bit of observer bias going on with your comment, as many popular sites these days run React and they happen to run like crap due to the reasons I listed above.

Hmm, does Asahi Linux translate between x86 / x64 and ARM layers (e.g., similar to Rosetta on MacOS)? If not, I can’t imagine it’s all that many.

(If Valve can somehow find a way to nail that, I imagine that’ll be a huge boon for portable PC gaming.)

A home address is considered PII (Personally Identifiable Information) because, when combined with a name, it directly identifies where someone lives.

When people share their address with businesses, they are generally protected under laws like GDPR and CCPA.

OP’s site seems a bit hosed at the moment so I can’t dig into their privacy policy and see how they handle GDPR and CCPA deletion requests.

Love the idea!

But! I don't love the idea of sending PII for friends and family to an unknown source in order to facilitate this.

You can ask your friendly neighborhood LLM for a template and go to town!

"I'd like to create a positive spin on "Cease & Desist" letters for the holidays that I could send to friends. Let's call them a "Continue & Persist" letter.

How can I make something cute and uplifting that I could send to friends for the holidays. I guess I need some personalization options like name, and one thing I'd like them to continue doing (optional).

Can you help me create an official looking, but fun form letter that I can fill in the send to people? Thank you!"

---

[Your Name or Family Name]

Department of Holiday Joy & Encouragement

[Your Address or City, if desired]

[Date]

OFFICIAL NOTICE

Re: Continue & Persist Order

Dear [Recipient's Name],

In the spirit of the holidays and in recognition of your outstanding contributions to the world, the Department of Holiday Joy & Encouragement has issued this Continue & Persist order for you.

This order officially requires you to continue being your amazing self and persist in spreading joy and kindness throughout the year. Please note the following specifics of your order:

1. You must continue to [something wonderful they do, e.g., make everyone laugh, inspire others, or bake the best cookies]. Your efforts in this area have been truly exceptional and have brought countless smiles.

2. Under no circumstances are you to cease being the incredible [friend/parent/artist/etc.] that you are.

3. This order is valid indefinitely and cannot be revoked (even if you try to protest with modesty).

Failure to comply with this order will result in an overflow of love, gratitude, and probably more letters like this one.

We at the Department of Holiday Joy & Encouragement commend your contributions to making life brighter and sincerely hope you have the happiest of holidays.

Warm regards,

[Your Name or Family Name]

Director of Holiday Cheer

That song was released in 1988 by NWA, a hip hop group from Compton, California.

I don’t think it’s too far fetched to think that song was colored by their experiences with the notoriously corrupt LAPD of the 1980s.

This is, of course, a non issue as the resolution of the Steam Deck is 1280x800.

Edit: I am silly. Of course, people mean hooking it up to a bigger screen.

I’m in the fortunate position of not needing to interview (I have a pretty solid thing going on right now), but I’m not opposed to going through the interview process if an interesting opportunity comes up.

So, I’ve done it a few times now (and ultimately been rejected every time, doh).

I tend to agree to OP's suggested alternatives.

For example, take home assignments:

Personally, I enjoy the take home assignments. It gives me time to work on something without pressure. I generally find them kind of interesting to do as well. The time estimates are just completely ridiculous though (e.g., "it should only take you 2 hours") and trying to work on them around normal work schedule, family obligations, and what not can be such a chore.

One company I interviewed with based one of the in-person technical interview panels on this take home assignment. We walked through the code, and even talked about things that could have been done better, given the constraints. Had a few other standard panels with them: system design, culture questions, etc.

Rejected. Doh.

Another company I interviewed with had a few pair programming exercises (these were virtual, in lieu of an in-person interview at the time) -- though they threw me for a loop. I've sat on interview panels myself, so, similar to OP, I feel like I have a good sense of what we should be looking for when we hire a candidate.

Oh, man. That does not translate well through the interview process. In my previous example, this company called me back for a follow up technical interview, because I "spent too much time communicating my development strategies and plan, and not enough time actually solving the problem." So, they called me back for an additional interview step. This time, I solved the problem without trouble and even had some good, engaging conversations as we went along.

Rejected. Doh.

The Leet Code stuff is just atrocious though. It's rote, it's not helpful, it often relies on stupid tricks and gotchas, and if you don't solve it in the most optimal way, you're pretty much done.

(Side note: Advent of Code is coming up in about a month or so and I am so looking forward to it. I'm serious when I say that those are FUN programming questions!)

Ultimately, I don't know what the solution is. We as engineers universally hate this process, yet we keep putting each other through it, finding the most minor things to ding a potential candidate for.

The reason that ChatGPT and other LLMs use such language is because they often find it in the corpus of text they’ve ingested. It’s a frequent pattern in writing.

So, I don’t necessarily think that’s an immediate red flag. To be sure, I guess you’d need to go back and look at the author’s previous stories and compare writing styles.

In 2015, we took the California Zephyr from Berkeley to Chicago and then (I forget which train) from Chicago to Washington D.C. The connection was pretty smooth. Met a friend in Chicago for drinks for a few hours while we waited for the DC train to depart.

I remember loading up ebooks, podcasts, etc, just in case I was “bored.”

No joke, I sat in the parlor car staring out at the landscape for 3 straight days!

It was absolutely glorious.

One fun thing (“fun”, depending on your perspective) is that you’re paired with random strangers at dinner.

The first night, I am telling the couple across from me how “awesome and civilized” this felt and it was nice not to have to deal with the bullshittery of TSA agents at airports.

They politely coughed and I knew instantly I screwed up.

“Yeah, we’re actually both TSA agents.”

Dinner was pretty awkward after that!

Fun fact:

Fat Mike once owned this swanky house in San Francisco [1]. Not too bad for “not having a job!”

Also, the dude founded and ran Fat Wreck Chords in 1990! He continues to run it with his partner to this day [2].

[1] https://www.sfgate.com/realestate/article/sf-home-punk-rock-...

[2] https://en.m.wikipedia.org/wiki/Fat_Wreck_Chords

Interestingly, they did this for Stable Cascade as well. [1]

[1] https://huggingface.co/stabilityai/stable-cascade/blob/main/...

How are you getting around Stability AI's Stable Diffusion 3 license restrictions? If I recall, it explicitly forbids providing access to the model through your own service / API. [1]

> You may not use the Software Products or Derivative Works to enable third parties to use the Software Products or Derivative Works as part of your hosted service or via your APIs

This is something we think about a lot on the open-sourced AI Horde distributed network -- we can't host this model due to these restrictions. [2]

[1] https://huggingface.co/stabilityai/stable-diffusion-3-medium...

[2] https://aihorde.net/

We do this with Gmail, too! Just make sure you log into your kids’ accounts every so often, otherwise Google is liable to wipe the account. [1]

[1] https://support.google.com/accounts/answer/12418290?hl=en

21 years, here! This made me go back to find my first post[1] from August 2003. Oof.

It’s interesting how things have evolved over time. Started it in college, and it was random posts about activities, feelings, etc. A lot of more personal stuff.

Growing up, changing careers (and changing again and again), having kids… I find that my blog has gotten less personal and more just general knowledge sharing of things I find interesting.

[1] https://daveschumaker.net/should-i-join-the-masses-of-blogge...

You have to strike while the iron is hot! Even iTerm just released an update that adds an “AI concierge” to your terminal…

And no, it’s still not April 1st!

Hah. I'd like to think there's nothing to fear from me as a user.

Look, I get why some of these policies are in place -- a bunch of it stems from locking down our systems and protecting critical data due to various Sarbanes-Oxley requirements. Plus, sometimes smart people do dumb things, and it leads to bad things (e.g, see the LinkedIn incident) [1].

But man, oh man, is it annoying! Especially if I'm in my own home, with no one around, and I otherwise get my work done.

[1] https://darknetdiaries.com/transcript/86/

Ah, that's an interesting point. I haven't tried to correlate that, but it must be true. For example, the screensaver never seems to appear during Zoom calls!