I like to call it "Exposing yourself to luck." You can't win lottery if you never play. Play the lottery with the best odds and play it as much as you can.
I think it’s just trying to be exposed to as many opportunities as possible with the best tools you can have. You can’t determine whether or not you can catch a fish, but you can acquire the best fish-catching tools you can get, try to get to the fullest river, and stay out there as long as you can. I think the role of “luck” really hits when you realize how some people are born on the side of the river with a gigantic, well-made net.
I’ve been using Zed Agent with GitHub Copilot’s models, but with GitHub planning to limit usage, I’m exploring alternatives.
Now I'm testing Claude Code’s $100 Max plan. It feels like magic - editing code and fixing compile errors until it builds. The downside is I’m reviewing the code a lot less since I just let the agent run.
So far, I’ve only tried it on vibe coding game development, where every model I’ve tested struggles. It says “I rewrote X to be more robust and fixed the bug you mentioned,” yet the bug still remains.
I suspect it will work better for backend web development I do for work: write a failing unit test, then ask the agent to implement the feature and make the test pass.
Also, give Zed’s Edit Predictions a try. When refactoring, I often just keep hitting Tab to accept suggestions throughout the file.
It feels like magic when it works and it at least gets the code to compile. Other models* would usually return a broken code. Specially when using a new release of a library. All the models use the old function signatures, but Claud Code then sees compile error and fixes it.
Compared to Zed Agent, Claude Code is:
- Better at editing files. Zed would sometimes return the file content in the chatbox instead of updating it. Zed Agent also inserted a new function in the middle of the existing function.
- Better at running tests/compiling. Zed struggled with nix environment and I don't remember it going to the update code -> run code -> update code feedback loop.
With this you can leave Claude Code alone for a few minutes, check back and give additional instructions. With Zed Agent it was more of a constantly monitoring / copy pasting and manually verifying everything.
*I haven't tested many of the other tools mentioned here, this is mostly my experience with Zed and copy/pasting code to AI.
I plan to test other tools when my Claude Code subscription expires next month.
Zed's agentic editing with Claude 3.7 + thinking does what you're describing testing out with the $100 Claude Code tool. Why leave the Zed editor and pay more to do something you can run for free/cheap within it instead?
I don't care about speed. I like how convenient is it to organize and move windows around compared to floating window managers. Especially when using multiple monitors.
My hand gets tired if I use mouse too much especially with dragging operations.
Using TWM and modal text editor helps a lot.
I really miss it on macos when I have lots of stuff open and have to switch between multiple windows. I haven't yet tried any tiling apps on mac because configuring is always a deep rabbit hole. I use Rectangle but having more than 1 window on a 13" screen isn't that useful and still requires manual arranging. A better control over workspaces is what I miss.
We used to manage 500+ servers with Ansible for almost 10 years. It was a nightmare.
With so many servers Ansible script would ocassionally fail on some servers (weird bugs, network issues, ...). Since the operations weren't always atomic we couldn't just re-run the script. it required fixing things manually.
Thanks to this and emergency patches/fixes on individual servers, we ended up with slightly different setup on the servers. This made debugging and upgrading a nightmare. Can this bug happen on all the server or just this one because it has a different minor version of package 'x'?
We switched to NixOS. It had a steep learning curve for us, with lots of doubts if this was the right decision. Converting all the servers to NixOS was a huge 2-year task.
Having all the servers running same configuration that is commited to GitHub, fully reproducable and tested in CI, on top of automatic updates of the servers done with GitHub action, was worth all the troubles we had with learning NixOS.
I realize Ansible is kinda slow and can be flaky, and wouldn't use it for 500 servers. However, for one beginner VPS I think it's fine.
The fact that it's not hermetic and perfectly reproducible is a major problem for a fleet, but for single user it's a benefit. It offers a graceful migration path from a snowflake server to a managed server, and still works even if you can't manage to do 100% of the config automatically.
I did the same thing when my grandfather died: A grid with each square representing a week of his life, and each row representing one year. Then, we mapped as many events from his life as possible.
Oh no, I don't go in that details, simply, every month, the weeks start on 1st, 8th, 15th, 22nd.
Only Jan, Jun, July, Dec gets another week, 5th one, on 27th.
So all months have 4 columns, labeled 1,8,15,22. Jan, Jun, Jul, Dec have 5 columns.
Whole sheet is a fillable grid of 52 columns and 90 rows.
A pre column list the year like 2002, 2003, 2004 etc. Next pre column list my age like 0,1,2,3 till 90. Then above explained grid. Then again, first two pre columns copied.
Each cell has light grey border. Border darkens to black between months columns. Border also darkens at every decade row.
If an event (like travel) happens on certain date (like 13th), it simply fills the whole week.
I assume that in overall big picture of life, a rounding off of week will not matter much.
Inkplate[1] might be a good option for around 100€. Compared to Joan it's less polished. You can get 3d printed case and it requires some coding (it doesn't support rendering HTML).
I'm working on a personal dashboard for the screen. I'll create a website in React and then use rendertron[2] to get the screenshot on the Inkplate.
