Adrien Baron
banner
abaron.net
Adrien Baron
@abaron.net
200 followers 620 following 140 posts
Software Engineer and Theatre fan living in London Currently building @theatreninja.com, a new theatre ticket platform for Londoners Maker of @clashofstats.com and http://tiny-frontend.github.io Working at @nagomy.security
Posts Media Videos Starter Packs
Reposted by Adrien Baron
Vitest 4 is out!

- Browser Mode is Stable
- Visual Regression Testing
- Improved Debugging
- Pool Stabilization
- New APIs
- Bug Fixes

Stay updated with our blog post:

vitest.dev/blog/vitest-4
Announcing Vitest 4.0
Vitest 4.0 Release Announcement
vitest.dev
I think the issue with "I can't change all in browser" is more the fact that in browser you can't edit components as a unit once they're rendered to the DOM.
Agreed, it's simply moving the composition paradigm, if you use a component framework, it already provides composition, and you don't need styling to provide that too (it's even detrimental as you end up with two separate composition trees).
If you don't use components, then CSS composition can help
Reposted by Adrien Baron
Guys new scamming method on bluesky. Please share this around. Dont fall for this stuff

Block, move on.
Each show gets it's own DO, which seems to be enough for now 😅, might have to find a way to scale/replicate popular shows at some point.
Alongside streaming, compression and LDJSON it unlocks cool things though, I can stream tickets for 400+ performances with thousands of tickets each from DO SQL to a browser in seconds.
Extreme example: www.theatreninja.com/musical/wick..., network payload is 6.5mb, uncompressed 126mb 😅
Wicked Tickets | London | Theatre Ninja
Find the best tickets for Wicked in London on Theatre Ninja. Defy Gravity with Wicked London, one of the West End's most beloved musicals, witnessed by millions..
www.theatreninja.com
Yeah, D1 is the same by default, but it does have read replication to handle this 😅
I use all this for my site www.theatreninja.com which sells theatre tickets in London, so my traffic is 95% UK which solves that particular issue 😅
Best London Theatre Tickets | Theatre Ninja
Find the best seat for your money. Explore all dates at once for any show and easily find the best seats available.
www.theatreninja.com
I also use SQL in DO and that should give you much better results (closer to their real metal SQLite perf) however you won’t be able to measure it with performance.now as it’s not async 😅
The CF dashboard will give you the stats though!
Yeah you’ll get the latency between the worker and D1 this way, and depending on where D1 and the worker are that probably will account for most of the time.
Can you share your methodology ☺️?
I’m using D1 in prod and even if not as fast as your local, I get much better number than this (2.2M requests, 0.19ms P50, 3.89ms P99 on a DB that’s 880mb)
The bubble may pop, but I don’t think we’re going back to a time without AI coding, quite the opposite.
Is it an issue for juniors that want to join? Yes. How do we get senior if we stop training? I dont know.
Do I think that sucks, yes.
But I don’t think we can’t put the genie back in the bottle
Current AI does work like a multiplier, you need to understand what good looks like to keep it on track, and that’s definitely an incentive to learn for more junior devs. I have seen AI improve by quite a bit in the last 2 years (especially the Anthropic models), so not sure it will stay that way
However 😅, I do use AI extensively both at work and on my side project and it does really work/makes me more productive, so do my coworkers, and we’re all senior.
I agree that understanding how things work behind the scene is important, and I do think people should upskill.
I also agree that AI is an economic bubble, and a lot of these valuations will collapse.
Reposted by Adrien Baron
Very excited to be able to talk about something I've been working on for a while now - we're working with Commonwealth Fusion Systems, IMO the leading fusion startup in the world, to take our work on AI and tokamaks and make it work at the frontier of fusion energy. deepmind.google/discover/blo...
Google DeepMind is bringing AI to the next generation of fusion energy
We’re announcing our research partnership with Commonwealth Fusion Systems (CFS) to bring clean, safe, limitless fusion energy closer to reality with our advanced AI systems. This partnership...
deepmind.google
Also, the “mistake” tradeoff might not be worth it:
- if you mess up with declarative, you end up re rendering too much, not great but not too bad
- if you mess up with manual, you end up not updating when you should: which is way more damaging IMO
I had the same feeling, I get being tired of React reactivity system (it’s very clunky!), but IMO Vue reactivity is way more sane, and that model is much nicer than manual update at scale.
Reposted by Adrien Baron
I posted this last night cause I kind of wanted to bury it. I got cold feet about putting it out there.

embedding-space.github.io/sparse-netwo...

The subject is WHY neural networks work, and I think the answer I offer is kind of interesting. Maybe even a little correct, possibly.
Reposted by Adrien Baron
It wasn't asylum seekers who increased anyone's rent or energy bills, the rich did that.
That’s such a cool idea!
I've released my first Go application!

It's a (very basic) terminal video editor that uses ffmpeg and whisper to let you selectively choose which lines from a video you want to keep.

You can also preview each line with a pop-up video player!

It's called tsplice:
Reposted by Adrien Baron
By the way, this is shockingly better than Discover. You should use it
Welcome to the ✨For You✨ feed!

It finds people who liked the same posts as you, and shows you what else they've liked recently.

📌 Pin to add it to your top bar
❤️ Like the feed and repost to spread the goodness