Sounds like a perfect job for @jason.energy & @codetv.dev 👀

Interesting, the hallway track is always the best part of conferences, but the talks are easier to "broadcast" with recordings, without that the conference is essentially invisible to non-participants.

A way to capture the organicity of hallway track discussions would be nice for reach post-conf.

Thank you! 🙌

OMG, TIL, thank you so much! 🙌

I tried lazygit yesterday, getting slowly pilled on it, but this is the cherry on top: it would also have prevented this 🙌

Archive as the website got the HN hug of death:
web.archive.org/web/20251111...

Would it have helped with my problem? (switching branches with uncommitted changes that got dropped)

This would be nice to pair with the try operator, but its return array being always truthy would maybe not do what you'd expect.

github.com/arthurfioret...

That's the mistake I made a couple of times with other solutions (Mirage comes to mind) and MSW: replicating the business logic behind the API/service in mocks.

When just mocking the IO is all you need, and any "smartness" is usually a sign of a test with too wide a scope.

The real 🤯 moment with MSW was when I understood you don't need to mock all the use-cases at once: you can just import the handlers (or better, define them at test setup time) you want to test against.

And it just works. 🙌

My reaction exactly.

I just read it, it is beautiful.

How does it handle changes in the source code when you already have translations? It detects a change in hashes and goes to to update them in each file?

What is this magic? 👀

This makes me so happy 🥹

No more patching the History API to detect navigation 🙌

And to add to the burden, AI review tools often make the problem worse here.

LLMs are noise machines, and code review is about cutting through the noise to find the right signals.