iiinova — where your digital self keeps evolving

Jakob Norlin @norlin.dev · 07/10/2026

We wrote up how we do it on our blog: endform.dev/blog/playwri...

Jakob Norlin @norlin.dev · 07/10/2026

With the Endform MCP server, the agent can access history and trace data from your previous test runs. So it can tell what type of issue it is, whether it's worth rerunning, or quarantine the test directly if it's been causing a lot of trouble lately.

100

Jakob Norlin @norlin.dev · 07/10/2026

The data needed to make these calls is usually buried deep in CI logs that are hard to access and don't paint the whole picture.

100

Jakob Norlin @norlin.dev · 07/10/2026

Here's one from this week. 1 failing test. It has failed 3 times in 1,038 runs over the last 4 weeks, with the same error each time, and the commit only touched an unrelated email file. Verdict: recurring flake, rerun.

100

Jakob Norlin @norlin.dev · 07/10/2026

Lately we've set up an agent to review every test failure in CI. It decides one of three things: → it's a real failure → it should be rerun → it's been flaky enough to quarantine

100

Jakob Norlin @norlin.dev · 07/10/2026

Usually there's way too much to dig through to figure out whether a failure is real, or just a flaky test, check or service. So my default reaction is to rerun without checking what's wrong.

100

Jakob Norlin @norlin.dev · 07/10/2026

How many times did you press "Re-run failed jobs" this week? I haven't been coding much the last few days, and I still think I'm at 5 😅

100

Jakob Norlin @norlin.dev · 07/09/2026

Endform allows you to limit concurrent tests, which solves both these use cases. Configure a tag that you want to limit concurrency on, what the limit should be, and whether it should only apply within a specific run of your test suite, or globally across all runs happening at the same time.

111

Jakob Norlin @norlin.dev · 07/09/2026

What if the reason for the lock is limited performance, and we actually want to allow 5 out of a 100 tests to run at the same time for a specific lock? Playwright does not allow you that level of control.

110

Jakob Norlin @norlin.dev · 07/09/2026

What happens if you have several CI jobs running at the same time accessing that same shared resource? Playwright has no way to know there are other test suites running, so will allow them, which risks tests mutating data relied upon by other currently running tests.

110

Jakob Norlin @norlin.dev · 07/09/2026

Playwright v1.63 was just released, and it has a new feature called "Test locks". This means only a single test with the same lock name can run at any one time. Great for tests that mutate shared resources that can not reasonably be generated uniquely for each test, but it has some limitations.

110

Jakob Norlin @norlin.dev · 09/08/2026

Understandable! It was just so spooky you all three stopped at about the same time

010

Jakob Norlin @norlin.dev · 09/08/2026

What about @prune602.bsky.social and @delfoo.bsky.social? You guys all stopped posting at the same time, and you were my whole feed 😅

170

Jakob Norlin @norlin.dev · 29/07/2026

In Scandinavia, Moomin merch is top notch merch!

250

Jakob Norlin @norlin.dev · 18/07/2026

Now called Doomsday Airport

040

Jakob Norlin @norlin.dev · 16/06/2026

Last week we shipped experimental Firefox support in Endform. This means you can now use the fastest Playwright test runner against twice as many browser targets! We're working on WebKit support to cover the whole suite of available browsers. endform.dev/changelog/20...

endform.dev

Experimental Firefox support - Changelog

Experimental Firefox support

000

Jakob Norlin @norlin.dev · 12/06/2026

That's not a bad idea. Any ideas for names?

000

Jakob Norlin @norlin.dev · 09/06/2026

Playwright is a great tool for keeping track of perf so you don't regress on perf metrics. You can integrate it with lighthouse and make sure the INP score is good for the interactions you want to stay fast. E2E tests are super versatile and can be used to test so many aspects of your application.

010

Jakob Norlin @norlin.dev · 09/06/2026

This is a fantastic read on how Linear keeps their application so fast performance.dev/how-is-linea.... We've already implemented a couple of these techniques in Endform, but we still have some ways to go.

performance.dev

How's Linear so fast? A technical breakdown

breakdown of the architecture behind Linear's speed: local-first sync, MobX observables, instant first loads, and a keyboard-first design.

100

Reposted by Jakob Norlin

Vite @vite.dev · 04/06/2026

Today, VoidZero joins Cloudflare. Vite remains MIT, vendor-neutral, and stewarded by the same wider team. The same goes for Vitest, Rolldown, and Oxc. Cloudflare is also committing $1M to an OSS fund to support independent development in the Vite ecosystem.

vite.dev

Cloudflare supports Vite's mission

The VoidZero team is joining Cloudflare. Vite remains MIT, vendor-neutral and stewarded by the same wider team.

521238

Jakob Norlin @norlin.dev · 02/06/2026

Love it when SolidJS gets more funding!

031

Jakob Norlin @norlin.dev · 01/06/2026

"Applies to jobs on my behalf" is one of the top promoted templates on Browserbase 😫

010

Jakob Norlin @norlin.dev · 30/05/2026

Example: “Use endform query to find failed checkout tests from the last 14 days, group them by error message, look for patterns across projects and branches, and suggest what in the test or app code is likely causing it.”

000

Jakob Norlin @norlin.dev · 30/05/2026

We just shipped endform query, a new CLI command for pulling Endform analytics data. A neat thing about this is that you can now ask your coding agent to help debug tests.

210

Jakob Norlin @norlin.dev · 29/05/2026

You don’t need to move your whole test suite into CI at once. Start by tagging up a few tests with "@ci" and then run "npm run playwright test --grep=@ci" in your CI pipeline. Tag up more tests as you go to increase confidence in what you ship.

010

Jakob Norlin @norlin.dev · 28/05/2026

2.5 megawatts of server rack space in MagasinX in Stockholm. Completely empty right now. What will they put in here? 🤔

000

Jakob Norlin @norlin.dev · 28/05/2026

Hello Mr Unitree robot 🤖 First one in Europe!

000

Jakob Norlin @norlin.dev · 17/05/2026

Or let others do it for you like in this case I guess

000

Jakob Norlin @norlin.dev · 17/05/2026

Keep your CI fast people!

100

Jakob Norlin @norlin.dev · 13/05/2026

So many v2 versions of the great tools we're building with brewing atm. Looking forward (with dread) to all the upgrades! @solidjs.com @drizzle.team alchemy.run are the ones I can come up with on the spot, but I'm probably missing one or two.

110

Reposted by Jakob Norlin

Ignacio Zsabo @izsabo.bsky.social · 29/04/2026

Zynth is a fine-grained reactive runtime for iOS and Android built on @solidjs.com principles. It enables an authentic native feel and smooth animations without blocking the JS thread, all from one codebase. Zend is now on the App Store. Open alpha: github.com/zynth-org/zy...

github.com

GitHub - zynth-org/zynth: Create cross-platform app using SolidJS

Create cross-platform app using SolidJS. Contribute to zynth-org/zynth development by creating an account on GitHub.

1184

Jakob Norlin @norlin.dev · 24/04/2026

Why is there so little content like this? I love seeing posts explaining problems solved, especially in the UI layer!

000

Jakob Norlin @norlin.dev · 13/04/2026

Yeah! Although we have a read-heavy dashboard, not many mutations, so Formisch-usage is very light

010

Jakob Norlin @norlin.dev · 13/04/2026

That's pretty much exactly our stack at @endform.dev, except for Hey API. The ecosystem has gotten so good!

110

Jakob Norlin @norlin.dev · 18/02/2026

XAMPP! Didn't change the default login for phpmyadmin. You can never guess what happened next!

140

Jakob Norlin @norlin.dev · 28/01/2026

Cool! I wonder what the fallback for that looks like 🤔 (not at a computer)

110

Jakob Norlin @norlin.dev · 01/01/2026

Yeah I agree. It was just a light hearted joke 😊 Although one that might have been pulled a few too many times

000

Reposted by Jakob Norlin

alex @astahmer.dev · 16/12/2025

technically it's been ready for mainstream adoption for a while now, since those were out and mostly stable: - tanstack (query / router / start / form) - ark-ui (ui lib) - shadcn distribution (copy/paste/dl from registries and adapt) so whats missing?

051

Jakob Norlin @norlin.dev · 25/11/2025

My LLM autocomplete just suggested the TS native type Exclude<...>. Didn't know about that, and it seems to be exactly what I needed. Almost like your DistributiveOmit, except it doesn't enforce "K extends keyof T", so it doesn't work for objects, like in your example. But good enough for unions!

140

Jakob Norlin @norlin.dev · 25/11/2025

Right on time! I just had to replace an omit with a manually specified subset of a union yesterday. I was very surprised the omit didn't just work. Will use this helper today to get some dignity back! 😆

130