Working on my first foray into DML-based speakers. Low expectations, but more enjoying the fun of learning a new domain (and it’s a distraction from building garden furniture). Got various exciters, panel types, mount ideas ready, just taking the time out of the evenings.
Google Sheets add-on for quickly generating forecasts on seasonal, time series data: https://www.youtube.com/watch?v=nPUdTCxURgg . Submitted it to the Google Workspace Marketplace a few days ago, hoping it'll get approved soon!
Sounds really cool! Do you have a page describing your project with some pictures? It's funny I checked your post history before asking and found that you made the same comment I'm doing today to someone doing something similar one year ago! Is it what got you started?
Not yet! I'm a mechanical engineer by trade, so my C++ is horrendous. I am still in the very early phases of this project, once I get past the hardcoding all of my variables I will share some more.
It can easily be a huge waste of time while still feeling like it's enriching. And even if it is enriching, if it gets in the way of other stuff you really need to to be doing, then it's a problem.
Hit the nail on the head. I've quit every other social media site with the exception of HackerNews. And while some would argue HN isn't social media, I would disagree; I'd argue most of what makes HN so popular is the commenters.
I've found that in this regard, HN is no less addictive to me than a site like Reddit. I impulsively check it when I'm bored and find myself wasting time in long comment threads that over the years I have found more and more pessimistic. Still far better than Reddit, but the pessimism gets to me to a point where I can't help but feel jaded. Anyway, I just don't think it's good for me anymore.
If you find you're spending too much time on hacker news, you can actually set anti-procrastination settings in your profile. I use them and it's definitely helped a lot!
I'm still pretty early in terms of figuring out if this approach is better for larger projects. I can confidently say it works great on small things - for small changes on small files the AST approach works pretty well. You can say things like "add a click listener to the button that calls a function to tally the user's score" to a simple game and it will do it. That is, less than a minute after typing the prompt, you will see the code update in the editor and see your preview-webview update with your changes applied.
However, I have noticed that the AST code quality heavily depends on how common it is in the training set. I think I will have to add documentation to it through RAG or something - because OpenAI's models that I'm using seem to have limited experience writing esprima for JavaScript for example.
So it's hit or miss. In some cases I do feel like I'm throwing stupid compute at solving small problems and it's unnecessary - however, as I work on the project, it is getting better and better at successfully making the modifications. Some of that is me improving the prompts, some of it is OpenAI improving the models themselves, and some of it is the infrastructure I'm building for the project itself.
I did notice a huge improvement when o1-mini released. It is dramatically better at writing the AST code than GPT-4o or 4o-mini. I haven't tried Claude 3.5 yet but I've been hearing it does an exceptional job at code writing - not sure about my AST requirements though!
Working on https://www.brief.news - a completely personalized daily newsletter on the topics you're interested in. We've just launched the ability to add custom topics, so you can create a newsletter on anything now!
Building a narrative interactive novel/RPG thing. Wrapping music composition this week, writing about half done, game engine built, art finishing around Christmas.
CV pipeline to get some additional realtime stats for an annual Mario Kart 8 LAN tournament my friends and I run, hoping to be able to get real time race position tracking and other stats like boost/drift %, per-player item distributions, average time to 10 coins, and whatever else we can think of (and make work)
You're going into first place too early. You kinda have to cruse in 2nd until the very end to avoid blue shells. Better yet, stay in 9th as long as possible - that's the highest position you can get the good drops like Blue Shells, Bullets, etc. But I think you can get red shells in 2nd, so just save those till the end.
I am working on (yet another?) DuckDB gui, with an emphasis on devops-y/back office workflows.
I want to join data from AWS with other sources and present a nice data-table UI, and perhaps allow taking some basic actions on a row, defining some filters, etc.
Have you ever tried to copy-paste data out of the AWS console? Truly a terrible experience.
Gnosi: https://www.gnosi.ai/ . Y
You can auto-generate a constantly updated encyclopedia from a set of documents and conversations with an AI about those documents. Trying to make the process of having a Zettlkasten or personal notetaking system frictionless.
Yeah! I am a computational neuroscientist and I was trying to understand this paper: https://arxiv.org/abs/1906.01678 so I had added the paper to my personal Gnosi and was asking questions about it. I had added this paper https://pubmed.ncbi.nlm.nih.gov/34512508/ to my Gnosi a few months prior after reading through it and completely forgotten about it. As I was asking questions Gnosi started making connections to this earlier paper and helping me understand the new paper in the context of this older paper I had already read and understood, as well as giving me some ideas on using thermodynamic neural networks as a toy model for how Parkinson's can be created by poor management of entropy.
Understanding something new more easily and making new connections and coming up with new ideas is exactly why I have worked hard on maintaining a zettlkasten for the past 6 years, but Gnosi allowed me to make the same kinds of connections much more frictionlessly.
We have similar success stories from our alpha users which include folks interested in anthropology, complex systems, UAP studies, physics, philosophy understanding new tech for startups, etc.
I’m working on a replacement for file coreutils (touch, cp, mv, rm) which prompt before deleting by default and have a more sane api overall. Intended for interactive use on the CLI. In Rust.
On writing some blog posts about things I’ve built lately (both at work and in my own time). Helps a lot with diving more deeply into topics than what’s reasonable for a “just needs to work” implementation.
Yeah, the rashomon effect was named after the movie. Actually I'm trying to update that particular article with data so check back again in a few weeks.
Also, as an artist, I love the domain name for your sites :)
A basic CRUD app around goal setting as a test for how AI tools can help write web apps to see how a modern web team could leverage this stuff to go faster, and maybe identify some missing pieces I could build one day. ChatGPT, Claude, Cursor, Copilot etc are genuinely great if you can already write code. They really let you blast straight through the mundane bits and focus on the hard stuff.
It seems that adding one node to a linked list causes a repeatable performance drop. That is, linked lists of say 2^21 nodes sort faster than lists of 2^21 + 1 nodes.
I don't think so. There's no garbage generated during the sort, which is what I'm timing. I'm sure that GC runs during the sorts, but it won't do significant amounts of work because of that.
A C-transliteration of my original Go algorithm shows similar overall performance, with performance drops at the same list lengths.
1) at/around powers of two
2) regardless whether using GC or preallocating
makes me think its got to do with cache sizes. For example, the M2 macbook has around 16MiB of cache, which gives you approximately 2 million ~ 2^20 nodes where each node has a 64 bit number and a 64 bit pointer.
The jumps in your measurements could be related to the cache hierarchy. On Linux (probably macOS too) you should be able to run "perf stat ./mergesort" to show you the hitrate of your CPU's caches.
Thanks for reading my blog post and thinking about it. I thought of cache as well, the 3 machines I'm running benchmarking on are all very different. The M2 Macbook has 128-byte cache lines, as opposed to the other 2 machines, which have 64-byte ache lines. The old Dell R530 has level 1 through 3 caches, the others have only levels 1 and 2. I'm still trying to understand CPU caching and see if I can correlate it to the performance drops somehow.
The linked list nodes are 16 bytes, and both 8-byte fields (.Next and .Data) get read and written during sorting. Up to a point, larger node sizes,which I would thing would change caching, don't change where the performance drops occur: https://bruceediger.com/posts/mergesort-investigation-7/
Is the compiler, or a library you're using, somehow adding code that takes a different strategy at runtime when the linked list gets bigger than the performance-fall-off size?
Good question, but I don't think so. Two mergesort variants show it, one does not. A C transliteration of one of the variants that show it, also shows performance drops. The Go compiler for MacOS on M2 silicon probably generates very different code than the X86_64 on Linux, yet they both have the same performance drops.
The code only uses standard Go packages. I wrote the linked list struct, I'm not using a package's struct or code. I don't even use standard package code during sorting, only in timing and list preparation.
What you suggest may be true, but it seems very unlikely.
1,062 comments
[ 0.24 ms ] story [ 410 ms ] threadI'm happy with boring old full-range.
In short, it is an asset management system tailored specifically for collectors.
Just added MFA via authenticator apps + recovery codes today.
I've found that in this regard, HN is no less addictive to me than a site like Reddit. I impulsively check it when I'm bored and find myself wasting time in long comment threads that over the years I have found more and more pessimistic. Still far better than Reddit, but the pessimism gets to me to a point where I can't help but feel jaded. Anyway, I just don't think it's good for me anymore.
My initial goal is to let users make webapp prototypes and iterate on them by writing tickets for the AI to complete.
I for some reason call it Code+=AI: https://codeplusequalsai.com
For example they're oh-so keen on doing math themselves, even though they're shit at it and I instructed them not to do any math.
It's also hit and miss if they implement the right method or not, even with low temperature.
In my case I was experimenting with translating simple word problems into matlab scripts, so a resultcould be computed.
Do you find the AST approach helps? Or is it mostly just throwing compute at it, ie larger more better?
However, I have noticed that the AST code quality heavily depends on how common it is in the training set. I think I will have to add documentation to it through RAG or something - because OpenAI's models that I'm using seem to have limited experience writing esprima for JavaScript for example.
So it's hit or miss. In some cases I do feel like I'm throwing stupid compute at solving small problems and it's unnecessary - however, as I work on the project, it is getting better and better at successfully making the modifications. Some of that is me improving the prompts, some of it is OpenAI improving the models themselves, and some of it is the infrastructure I'm building for the project itself.
I did notice a huge improvement when o1-mini released. It is dramatically better at writing the AST code than GPT-4o or 4o-mini. I haven't tried Claude 3.5 yet but I've been hearing it does an exceptional job at code writing - not sure about my AST requirements though!
- Doing a massive tech modernization for a global nonprofit.
-Ghostwriting educational email courses for agTech founders who want to convert more customers or investors.
- IF I'm good and get my chores done, I may let myself build a better way to apply for agTech grants.
Aiming to launch next summer, with various media.
Currently working on supporting dataview tasks format and multiple reminder lists.
The app supports any markdown backed notes app, but I fear I may have limited its appeal by including Obsidian in the app name.
Edit: previous discussion https://news.ycombinator.com/item?id=39764919
I want to join data from AWS with other sources and present a nice data-table UI, and perhaps allow taking some basic actions on a row, defining some filters, etc.
Have you ever tried to copy-paste data out of the AWS console? Truly a terrible experience.
We are building out the Vanta equivalent for sustainability and climate disclosures.
Understanding something new more easily and making new connections and coming up with new ideas is exactly why I have worked hard on maintaining a zettlkasten for the past 6 years, but Gnosi allowed me to make the same kinds of connections much more frictionlessly.
We have similar success stories from our alpha users which include folks interested in anthropology, complex systems, UAP studies, physics, philosophy understanding new tech for startups, etc.
The tech stack is .NET and OpenGL.
Progress has been a bit slower than I wanted mostly because I've been sick but we'll get to an MVP some day!
Recently, a fairly detailed one on doing something semi-obscure with directory services on AWS. https://excessivelyadequate.com/posts/sadwsp.html
Also, as an artist, I love the domain name for your sites :)
https://bruceediger.com/posts/mergesort-investigation-1/
It seems that adding one node to a linked list causes a repeatable performance drop. That is, linked lists of say 2^21 nodes sort faster than lists of 2^21 + 1 nodes.
A C-transliteration of my original Go algorithm shows similar overall performance, with performance drops at the same list lengths.
Forcing garbage collection right after each sort (releasing all references to the sorted linked list) didn't change anything, either: https://bruceediger.com/posts/mergesort-investigation-10/
The jumps in your measurements could be related to the cache hierarchy. On Linux (probably macOS too) you should be able to run "perf stat ./mergesort" to show you the hitrate of your CPU's caches.
I'd be interested in a follow-up post :)
The linked list nodes are 16 bytes, and both 8-byte fields (.Next and .Data) get read and written during sorting. Up to a point, larger node sizes,which I would thing would change caching, don't change where the performance drops occur: https://bruceediger.com/posts/mergesort-investigation-7/
Sorting an already sorted, or reverse sorted list, doesn't exhibit the performance drops either: https://bruceediger.com/posts/mergesort-investigation-9/
A purely recursive mergesort doesn't show the performance drops: https://bruceediger.com/posts/mergesort-investigation-3/
The code only uses standard Go packages. I wrote the linked list struct, I'm not using a package's struct or code. I don't even use standard package code during sorting, only in timing and list preparation.
What you suggest may be true, but it seems very unlikely.