Hashimoto argues Zig gave him seamless C interop for calling libghostty from Swift on macOS and GTK on Linux without fighting a borrow checker across FFI boundaries. He also wanted granular control over allocations that a terminal renderer specifically needs, and accepts the tradeoffs of a smaller ecosystem and occasional compiler breakage.
The editorial argues that in a category where Alacritty, Kitty, WezTerm, and iTerm2 are all racing to bolt on more surface area, Ghostty's deliberate refusal to add a plugin system, config-as-code, or IDE features is what actually distinguishes it. The interesting product question is what Hashimoto explicitly refused to build, not what he shipped.
The submitter surfaced the interview to a receptive HN audience that upvoted it to 173 points, with a recurring comment theme from senior engineers agreeing that terminals were overdue for serious craft attention. The reception validates Hashimoto's framing that the terminal was a neglected core tool despite constant use.
Alex Alejandre published a long-form interview with Mitchell Hashimoto — co-founder of HashiCorp, author of Vagrant, and, more recently, the person behind Ghostty, the native GPU-accelerated terminal emulator that shipped a public 1.0 in late 2024 and has since become the default for a noisy slice of the developer internet.
The interview covers ground that most launch-week coverage skipped: why Hashimoto chose Zig over Rust, how he thinks about scope for a project that has no revenue model and no roadmap obligations, and what he learned from a decade of watching infrastructure tools calcify. He is unusually direct about the tradeoffs: Zig gave him the C interop and manual memory control he wanted, at the cost of a smaller ecosystem and a compiler that still occasionally breaks his build.
The HN thread — 173 points and climbing — is worth reading for the comments alone. A recurring theme: senior engineers who agree with his framing that terminals had quietly become the most-used, least-loved piece of software on their machines.
Ghostty is not the first fast terminal. Alacritty, Kitty, WezTerm, and iTerm2 have all been chasing the same axes for years. What makes Hashimoto's approach interesting is what he explicitly refused to build. No plugin system. No config-as-code Turing tarpit. No attempt to become a terminal-shaped IDE. In a category where every competitor is racing to bolt on more surface area, Ghostty's product decision was subtraction.
The language choice is the part practitioners should actually chew on. Rust is the default answer for "I want a fast native app in 2026," and Hashimoto is not naive about that. His argument for Zig is narrow and specific: he wanted a language where he could call into libghostty from Swift on macOS and GTK on Linux without fighting a borrow checker across an FFI boundary, and where he could reason about allocations at the granularity a terminal renderer actually needs. Zig gave him that; Rust would have made him work harder for it. He is not claiming Zig is better. He is claiming it fit the shape of *this* problem.
The second thing worth noting is how he scopes the project. HashiCorp — the company he co-founded and left in 2023 — is a case study in what happens when a tool that started opinionated accumulates enterprise features until every deployment is a snowflake. Ghostty is his attempt to build software the way he wishes he had built Vagrant: with a hard perimeter, a native binary, and no plugin ecosystem to inherit responsibility for. The interview reads, at times, like a postmortem on a career's worth of scope creep.
Community reaction has been divided in a healthy way. The Rust crowd, predictably, thinks he underrates cargo and the crates ecosystem. The Zig crowd is delighted to have a marquee project. The larger group — engineers who don't care about the language debate — mostly wants to know if it renders their Nerd Font correctly and whether their tmux setup survives the switch. (Mostly yes to both, with the usual caveats.)
If you are still on iTerm2 or the default macOS Terminal, Ghostty is now a defensible switch. GPU rendering, native macOS integration, and a config file that fits on one screen mean the migration cost is measured in minutes, not evenings. The main reason not to switch is if you have deep muscle memory in a plugin-heavy setup (WezTerm's Lua config, Kitty's kittens) — Ghostty will not replicate that, and Hashimoto has said explicitly it never will.
The language lesson is the more portable one. Most side projects fail not because the author picked the wrong language but because they picked a language optimized for a different problem and then spent their weekends fighting it. Hashimoto's Zig choice is a reminder that "what's popular" and "what fits" are different questions, and that a senior engineer's edge is often just being willing to answer the second one honestly. If you are starting a native app, the shortlist is still Rust, Swift, Zig, and (grudgingly) C++. Pick based on FFI shape and memory model, not vibes.
The scope lesson may be the most useful part of the interview for anyone shipping a tool right now. Ghostty ships without a plugin system on purpose. That decision closes doors, but it also means the maintainers can actually finish things. If your side project is drowning in feature requests, the Ghostty model — draw a perimeter, defend it publicly, ship the core well — is a viable answer.
Ghostty's next act is Linux polish and Windows support, and Hashimoto is candid that the latter is a long way off. The more interesting question is whether Zig itself graduates from "Hashimoto's choice" to a broader systems-language contender before its 1.0 lands. For now, the interview is the clearest public statement of how one of the more thoughtful engineers of the last decade picks his battles — and it's a useful read whether or not you ever install the terminal.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.