The best thing about building a GPU-powered 'serverless' platform (based on Knative) is that I can play with it early, and deploy lots of things. I'm totally into GenAI now! Let's use all these GPUs ;-)

Great video of the last Dagger meetup at Ubisoft HQ youtube.com/watch?v=gCpBjQ… - can you feel the Dagger vibes? ;-)

I'm playing with TabbyML github.com/TabbyML/tabby and I just discovered that they are using tantivy. It's a small world...

the new 'github is down, I can't work' is 'huggingface is down, I can't work' ;-)

We are now publishing #Kubernetes KinD node images on the Flux Project GitHub Container Registry. If you want to try out Kubernetes 1.30 on Apple ARM or Intel you can use the images hosted on GHCR.
github.com/fluxcd/flux-be…

here is a little initial release. literally use flask! or any http framework: github.com/progrium/httpf…

If you deploy all things with Helm & Flux, you should consider prioritising Kubernetes 1.30 upgrade. Especially on clusters with lots of CRDs, the number of API calls Helm SDK does will decrease substantially.

Lots of awesome changes in @KubernetesIO 1.30! But my favorite feature isn't mentioned in the post: Mounting volumes in debug containers! Minimal base images are now an option!

Thanks to everyone involved and the tirelessly working release team! ✨

buff.ly/3Q8Y2gy

Great BBL at Ubisoft today with Etienne Menguy about Ceph!

Introducing Dagger Traces: debug and monitor your Dagger Functions dagger.io/blog/introduci…

We've been tuning bincapz to reduce false positives after scanning it across all the packages in Wolfi OS.

Give it a try, the results are getting really good: github.com/chainguard-dev…

This is level the of clarity that's going to democratize AI/ML. We need more people that can explain what's happening behind the scenes, in simple terms, if we really want this next wave of technology to benefit everyone, and not just a handful of corporations.

I've played with GPTScript_ai native OpenAPI integration, and... it's awesome! We have an internal platform composed of multiple independent services, and we can now ask questions that will spread across different services. Automagically ;-) docs.gptscript.ai/tools/openapi

📬Just published 'Boost your #API mocking workflow with ollama and Microcks'. Generate samples and test suite for your #API , #microservices and #EDA , the easy and local way😉

👉 link.medium.com/qDBaHTRktIb

This is explains how the xz backdoor was found

Such an incredible timeline on the xz hack, boehs.org/node/everythin…

I wonder what the distros can do to protect on attack targeting their build system with such level of obfuscation

it is amazing to me a postgres core contributor saw ssh latency go up for no reason, investigated and completely broke a ccp backdoor plot just as it was going into prod after 2 years of careful progress, postgres saves the world again

Andres Freund, the principal software engineer at Microsoft who discovered the xz backdoor really does deserve a big pat on the back. 👏

The outcome could have been much, much worse.

The xz backdoor is, well, setting a fire under the entire Linux ecosystem... but I'm also so impressed with how it was set up: 2-yr maintainership, oss-fuzz, etc.

...and who knows how long it would've stayed undetected if the injected sshd code ran faster (<600ms)

Highlights:

Dagger hit 10K stars on GitHub!

We couldn't have done it without our amazing community. Get a glimpse into everything our Daggernauts have accomplished here: dagger.io/blog/10k-githu…