`/feedback https://github.com/anthropics/claude-code/issues`

Is that good, or do I need a separate one for each of the 10,000 currently open issues?

(Just messing with you. The number of unaddressed open issues is frustrating, but it is nice of you to be here offering to help despite those)

I had the opportunity to tiredly stumble upon Mark's display on my way out of a pinball expo. The quality and care that went into the displays move them beyond a learning tool and curiosity piece into a creative artistic expression. There's obviously a human who cares deeply about the exhibit. In this case, he was sitting in the back of the room near a mountain of travel crates, offering occasional encouragement and insight.

The message that I got was, "This is important to me, and I want to share it with you". It was not built out of a CEO's metrics and mandates asking me, "How many dollars are you?". It helped that there was literally a human behind the display, but maybe that same sense comes across via the website as well.

Regardless, it was refreshing. I stayed and browsed the whole display. Thank you, again, Mark!

I emailed government employees until I figured out who was responsible for license plate records. I submitted a CORA (Colorado Open Records Request) for the entirety of their dataset. I had hoped to get the data on some regular cadence to build a simple online service for others. Unfortunately, they flat out refused and wouldn't discuss options.

When I told my family what license plate I wanted, they laughed at me and said "No one has that, just go get it". And so I did and it worked. I now have what I consider to be the best possible license plate in Colorado: "LCNZPLT"

Occasionally I'll see someone walk by my car, see the plate, think for a few seconds and then start laughing. Mission accomplished!

I was getting intermittent repeating characters on random keys after replacing my mechanical keyboard switches with Outemo switches (spoiler: they were not the issue). This was often outside of a text-input application, spamming Ubuntu, preventing input and causing freeze-like issues. In addition, I often didn't recognize it right away and could not reliably determine which recent modifier or other keys I had just been pressing. There was no physical indication on the keyboard to alert me either.

Solution:

I needed a way to clearly identify when a key was stuck so I could isolate other hardware components. I also needed to know which keys it was happening on to find out if it was a small set of bad switches or randomly any key.

I wasn't able to find an existing solution that 1. Monitored continuously 2. Checked for repeating keys rather than just single presses.

Implementation:

Built for Linux and tested on Ubuntu.

This runs as a simple bash script after the user configures their input device name. It can optionally be set up as a system daemon. (The README has comprehensive instructions)

The script: 1. Sets up an input stream from `evtest` 2. monitors for repeat events 3. Notifies and logs the key name when the configurable threshold is exceeded 4. ... and again if/when the same key gets unstuck

While the body of the script was designed specifically for my use case, ie repeat keys, the project is meant as a template that sets up monitoring and notifications/logging for physical inputs. You could change it to monitor other input streams, check for press down/release events, duplicate/bounce key strikes or limit the set of keys that's monitored.

Thanks!

I hope that somebody, somewhere finds this useful in some form, even if just as a reference!

Comments URL: https://news.ycombinator.com/item?id=46141629

Points: 1

# Comments: 0

Not specifically because of the AI perspective, but because of the broader recognition that we need better development tooling. I spend a lot of time thinking about how developers work, reason about and mentally model software. I was specifically excited because of the example "HUD" that Geoffrey chose to share: "A custom debugger UI which visualized the behavior of my program".

Yes, please, I agree. That type of code flow interface should exist everywhere. So *I built one*. Ruby is my pilot language, but more will come later.

Code Beacon is a VS Code extension that analyzes and displays tracing data from a Ruby tracer gem. At its core, it filters non-application traces and rolls up the raw trace to a much smaller call graph so you can easily see and navigate to what files/classes/methods got run and how they interact.

It's not finished at all, but it's a start and it should mostly work. If it doesn't and you want to let me know, I'll fix it. It doesn't work in containers, it's not in all marketplaces yet, it will fall over if your call trace exceeds 10MB, and it could be a lot more performant. The UI/UX needs refinement and I'm sure there are plenty of bugs and edge cases to address.

I would love to hear your thoughts!

Comments URL: https://news.ycombinator.com/item?id=44747111

Points: 1

# Comments: 1