
Mitchell Hashimoto
@mitchellh
Followers
140K
Following
21
Media
2K
Statuses
37K
Working on a new terminal: Ghostty. 👻 Prev: founded @HashiCorp. Created Vagrant, Terraform, Vault, and others. Vision Jet Pilot. 👨✈️
Los Angeles, CA
Joined January 2008
Ghostty's unit tests now fully pass Valgrind, which is a significant surface area particularly for core terminal emulation features. And we run Valgrind in CI on every commit! This is a really good extra safety net to have.
github.com
👻 Ghostty is a fast, feature-rich, and cross-platform terminal emulator that uses platform-native UI and GPU acceleration. - ghostty-org/ghostty
17
13
826
It's extremely frustrating to see complaints about tech non-profits being underfunded while they're also consistently the worst run non-profits I've ever seen. I want them to do better, so I wrote up some advice for technical non-profits:
mitchellh.com
3
8
184
Going forward, AI assistance must be disclosed for Ghostty contributions:
github.com
I think, at this stage of AI, it is a common courtesy to disclose this. In a perfect world, AI assistance would produce equal or higher quality work than any human. That isn't the world we ...
78
133
2K
RT @PardonMyTake: Tuesday night max woke Big Cat up with a flashlight at 2am because he thought we were going to get sued. @forthepeople ht….
0
11
0
We rewrote the Ghostty GTK application from scratch and verified every feature with Valgrind along the way. Here's a reflection, plus notes about memory safety and Zig from a complex, real world codebase used by many, many thousands all day everyday.
mitchellh.com
16
57
782
Agree. One shotting stops being practical at an early complexity level. From that point forward you have to start getting a lot better at collaborating on a spec and game plan with the AI and giving it the guardrails to succeed (how to build, test, run). Then let her fly. For.
The best way to learn to code with AI isn’t to try and “one shot” things. Start by micromanaging things at the tiniest level, asking it to do very specific tasks. It’s way easier to get a knack for how to get good results, and you can slowly go up in abstraction from there.
21
22
446