cyrilou242

Building a video game adaptation of Pass The Pigs (the "dice" game where "dice" are plastic pigs).

Early preview here: https://piggy-toss.netlify.app/

The goal is to play with friends, we love this game.

I’m building Streamlit for Java. https://javelit.io https://github.com/javelit/javelit

I maintain a dev log: https://world.hey.com/cdecatheu/javelit-diary-00-building-a-...

And here’s an article about the project by a Google Cloud devrel :)

https://glaforge.dev/posts/2025/10/24/javelit-to-create-quic...

Hey, I'm the creator of Javelit. I don't know who you are microflash, but thanks for sharing! Short-story: Javelit is a copy of Streamlit for Java.

Any question let me know!

> Everything you do needs to be synchronous and almost everything requires a reload.

this is pretty incorrect, you may want to look into the concept of "partials" in SSR. maybe you meant everything requires a roundtrip ? but SPA would not solve most of the roundtrips necessary in github given many interactions in the github app require authn/authz checks.

would you care getting into more details ?

Also, 'old' github was known to be very fast an reliable and was indeed a ruby on rails SSR app. Since a few years ago github started to introduce react and more client side logic and it correlates with more issues and more slowness in the frontend. It only correlates, but still.

Sadly the paper uses for benchmarks datasets that: - are known to be pretty useless - contain mistakes - can be misleading with a naive F1-score measure. (to be fair they write "we looked at the F1-Score, under which both partial and full anomaly detection are considered correct identification" so this may be mitigated, but it's not clear)

See https://kdd-milets.github.io/milets2021/slides/Irrational%20... So it's hard to take any benchmark from the paper seriously. The paper is also ignoring any recent work (like > 2018) on univariate timeseries anomaly detection in the matrix profile space (eg MADRID).

The "practicality of usage" and conclusion sections are pretty correct though: it's expensive, slow, and no-shot is worthless if some other methods can train and infer in orders of magnitude less time.

It would have been interesting to see how the DETECTOR method performs when the LLM forecasting is replaced with some standard forecasting. (eg some auto ETS, if possible robust to anomalies in the training data). It looks like the natural follow up of this article is to remove the LLM altogether.

Can't wait for a cancel article from the Amazon people now.

"Google is aware that an exploit for CVE-2023-2136 exists in the wild." https://chromereleases.googleblog.com/2023/04/stable-channel...