pickle-wizard

I just started experimenting with the decision models. I spun up Laya on a VM with a couple of vCPU and 6GB of RAM. I get the results in about half a second. No need for GPUs or tons of memory.

I am integrating it into the product I am building and to me it doesn't seem like there is much need to go with a SaaS for this since the requirements are so light. I just can run it in Cloud Run and get all of the scale I'll ever need, and I get to tell my customers their data never leaves my environment.

I hadn't heard of this project. However this did give me something to think about.

I am working on a product I intend to open source for 18 months now. One thought I have had is how to keep people from thinking I vibe-coded the whole thing. I have been working on it in a private GitHub repo. When I release I was planning to copy the final code into a new public repo. I probably shouldn't do that. Just make this one public and let people see the evolution of code, even as ugly as a process as it has been.

I am a solo developer. I write most of my code by hand. I use LLMs to help me do research and to help find stuff in documentation. I also use it to do come reviews since I don't have anyone else to review the code. I do get a lot of value out of the code review. It often finds little things I missed or will tell me if my code sucks.

Really for what I am doing writing the code is never the bottleneck. I spend little of my time actually writing code. Where I spend most of my time on the design and requirements.

Microsoft built this so they can have a MongoDB compatible database service for Azure.

A few years back MongoDB changed their license to AGPL to make it harder for companies to take their code and offer it as a service. They did this because it competes with the MongoDB Atlas service.

At a past job the security team wouldn't let us have netcat or curl on our systems. So I just used /dev/TCP to get around that. The ergonomics were not as nice as using netcat or curl, but it got the job done.

I am still using VSCode. I do have the Claude Code extension, but I don't really use Claude to write code. I mainly use it for code reviews. Since I am a solo developer it is nice to be able to get a second opinion.

I use a passkey that is in iCloud Keychain.

There is a big difference between vibe coding and using an LLM to assist with coding. Vide coding is just accepting what ever slop it outputs and saying LGTM. You are not going to get quality code that way. You need to review what it does. I often find that it uses outdated libraries and deprecated functions.

Some ways an LLM can assist with coding: I recently needed to refactor a bunch of code. Claude was very helpful this and it completed in about 5 minutes what would have taken me a couple of hours by hand.

Also they are very handy when using new frameworks and libraries. As we all know documentation for open source projects is often lacking. Just yesterday I ran into this. I pointed Claude at the projects GitHub repo and had the answers to my questions in just a couple of minutes. Manually I would have been spending a hour or two reading the code to figure out what I needed.

They are very handy when debugging. Get a weird hours that makes no sense. Instead of banging your head against the wall for a few hours, an LLM can help you find the problem much quicker.

I was discussing this with a friend last night. If fuel prices get high enough I can see it happening.

In a past life I was a Domino admin and I ran it on an AS/400. It was great, handled incredible amounts of load, and required little maintenance. The only outage we had was when I accidentally deleted all the databases from the file system. Even that was pretty easy to recover from.

What about the old school processors like Authorize.net and Cybersource? It has been over a decade since I worked with them, so not sure what they are like now.

It has been 25 years, but back in college I had a job refurbishing and repairing PCs. Most problems were caused by cheap no name hardware. Most the quality hardware rarely had problems.

On a lark a friend setup Ollama on a 8GB Raspberry Pi with one of the smaller models. It worked by it was very slow. IIRC it did 1 token/second.

If you don't need color I recommend going with a laser printer. Not only for lower consumable costs, but better software support too. In 2009 I bought a Lexmark laser printer new for $120 at Frys. It has native postscript support, so it will work with anything without having to load software from the manufacturer. I systems ranging from an old Macintosh Quadra running System 7 to new Mac running MacOS Tahoe, a PC running Windows 11 and everything in between. It even works in Linux too.

The consumables are cheap too. I have just replaced the toner cartridge once and a new OEM one was about $80 on Amazon.

The same way you handle any fraudulent charges. Notify the bank and they'll reverse the charges.