I've had at least one instance of Ghostty running on both my work and personal machine continuously since I first got access to the beta last November, and I haven't seen a single segfault in that entire time. When have you seen them?

Look at the issue tracker and its history too.

I've seen the amount of effort Mitchell &co put into ensuring memory safety of Ghostty in the 1.2 release notes, but after upgrading I am still afraid to open a new pane while there's streaming output in the current one because in 1.1.3 that meant a crash more often than not.

So Ghostty was first publicly released on I think December 27th last year, then 1.0.1, 1.1.0, 1.1.1, and 1.1.2 were released within the next month and a half to fix bugs found by the large influx of users, and there hasn't been a segfault reported since. I would recommend that users who are finding a large number of segfaults should probably report it to the maintainers.

Bun is much worse in this regard too.

It makes me sad, because they demonstrated JavaScriptCore is shockingly better than V8 for node-likes. The Typescript compiler (which like basically any non-trivial typechecker is CPU bound) is consistently at least 2x faster with Bun on large projects I've worked on.

When Typescript finishes their Go rewrite that will become irrelevant, and I rather have the compiler from the same people that design the language.

For that example sure, and admittedly the entire JavaScript/TypeScript processing ecosystem is moving in that direction. But the TypeScript compiler is not the only CPU-bound JavaScript out there.

There are plenty of memory safe compiled languages to rewrite that JavaScript into.

segfaults raise my belief in spirits

Possibly a good Halloween costume idea to go as a segfault. It would scare some people.

I haven't seen a single one.