Connect your Claude, Codex or OpenClaw for full transcripts on millions of podcasts. Two week free trial, cancel anytime.
Start free trial →Artisan Beer · AI Builders · November 2026
Ed was explaining his test-segmentation method: sorting a bloated test suite into tests to run on every push, tests to run nightly, and tests to delete.
…tests should run every time CI runs. The tests that give you real confidence that you didn't break anything. And then there's a whole other big bucket of tests that's tests that are good to have, but I don't need to spend 20 minutes waiting for CI to pass. So maybe we just run those every night.
Even a bunch of tests we can probably delete. Yeah.
And a bunch of tests. And then a tombstoning test. Well, I mean, I don't know. It's kind of hard to monetize it. I've been tinkering with trying to make some deterministic scripting to kind of make it easier. But a lot of times it's sort of a judgment call and it comes down to your own personal risk tolerance, like how many tests you feel comfortable deleting. Because the whole point to a test suite, right? It's to give you confidence that your application continues to work after you've made changes. And so how far can you par things down before you start losing that confidence? And I think that answer is going to be different for everybody. So I don't really want to make that decision by bundling it into a product. I can do is I can create like real planned scripted thought through educational content around that. And so there will be that sort of content will be in scalpels. And…
I did it with a headline right there.
Yeah. And I did it without really, I mean, there was a small reduction in confidence, right? There's, there are definitely things that are not being tested in CI that could break. But then when you combine that with the fact that I have all of these sites being monitored with Hone and Wake. And so…
And things in the places