I had a discussion with a friend the other day, and an example made there brought some clarity to my thinking about LLMs and how they’re currently being used in dev workflows.

A significant number of developers that I interview describe their dev workflow as, effectively, managing the LLMs as if they were a team of junior engineers. The developer does the planning and architecture work, perhaps with collaboration from an LLM agent, and then sends out the tasks to individual coding agents to do implementation. Finally, the results are aggregated and the developer reviews it all before signing off on it.

Over the years, I’ve become more and more enamored with static assets. Don’t get me wrong, there’s a place for dynamically generated content. But after spending 15+ years building webapps and websites, I’ve reached the point where I truly think the majority of the web is overcomplicated. Wordpress should die in a fire.

Harsh? Yes, probably. But the primary benefit that Wordpress provides is that it enables users to produce their own content and publish it immediately, rather than relying on a developer of some flavor to make the changes in code and push it up to the server. 15 years ago, that was revolutionary and the explosion of cheap Wordpress hosting is the proof in the pudding. Realistically, cPanel exists almost completely because Wordpress shattered the website industry and took over.

A quick note about uploading documents to Paperless NGX via the API: You need to make sure that your multipart form document part includes the filename attribute.

This is automatically done by a browser when it’s using a file field to pull the file, but if you’re constructing the multipart form manually, the API will reject an upload without the filename with the following message:

The submitted data was not a file. Check the encoding type on the form.

Heyo, just as a quick announcement, I’ve added a new page to the site dedicated to Pathfinding. My hope is to grow it into an entire section of the site, with detailed information about pathfinding algorithms, breakdowns and summaries of academic research papers, and links to other helpful sites with more information and demos, but for now it’s a giant braindump of information.

Of particular note is the table of papers at the end of the page. While not comprehensive, it does cover a number of major papers that have influenced the field and produced commonly-used algorithms. I’ve added some quick notes for each paper, as well as links to where you can view them without needing a journal subscription (one of the few things I miss about being a student; journal subs are expensive).

Document everything.

No, seriously. Document everything. You think you’ll remember what you’re doing right now. You think you’ll remember exactly how machine users should be set up in AWS, because it’s so obvious, how could you possibly forget it? You think you’ll remember exactly where you’re storing that system backup configuration, and how to decrypt it, because it’s the most logical thing you ever came up with. It makes perfect sense, Future You would be an idiot to not remember that.