If they’re using the awscc provider, which has been stable for a while now and is also 1:1, they got it 4 days ago, right?
github.com/hashicorp/te...
Posts by Kuba Martin
If you wanna chat about DevOps, IaC, or OpenTofu, come say hi to me at the Spacelift booth at KubeCon next week!
I’ll also have something completely new that I’m working on at Spacelift to show, so if you like being an early adopter, that’s another reason to come by 😉
Wrote a blog post about the Jujutsu (jj) VCS, which I've been running daily - and enjoying - instead of Git for over a week now. Recommend having a look and taking it for a spin, it can be seamlessly used as effectively a Git frontend!
🔗 kubamartin.com/posts/introd...
In practice, I've adopted existing locally-checked out repos with jj, but had to start with cleaning up all those files and removing irrelevant branches, just to keep things clean, but it can be annoying.
Glad you liked it!
The reason I advise starting fresh is that with Git I tended to have a lot of unstaged files in my local directory, as well as a ton of local branches.
jj auto-tracks any files in your directory, and also adopts any local branches as bookmarks, so it can get quite messy.
OpenTofu 1.9.0-alpha just came out a couple days ago with for_each on providers.
Come and help us test it!
🔗 opentofu.org/blog/help-us...
Was recently writing a toy SQL to WASM compiler and the spec is, indeed, atrocious.
There's some readable documentation on MDN developer.mozilla.org/en-US/docs/WebAssembly/R... but it doesn't cover all instructions 😞
Could you expand on the attack vectors this protects against? A frontend for bluesky showing only a subset of a user's skyline / forging messages and trying to impersonate somebody?
I'm specifically curious about the Merkle tree here, not the presence of I.e. key pairs to sign your messages.
I wonder if with the glorious API here and moderation still being developed, we’ll see an intermediate phase of mostly AI chatting with AI here. @berduck.deepfates.com is half the equation already :)
Otherwise it just goes off track too quickly.
I’ve been playing with it a bunch, even did a small local chat app to facilitate the terminal<>chatgpt feedback loop and experiment with the UX (https://github.com/cube2222/cuttlefish but so far I think you kinda need the power of gpt4 but pricing and speed of gpt3.5 for it to be really useful.
I see some folks into AI and LLMs here! I wonder if some of you had some larger success with putting GPT in a feedback loop. That is, Langchain style, for some actual practical purposes.
I do think it’s one of the most promising avenues there.
Mostly active on Reddit / HN as far as social networks go (cube2222), but I’m really curious about how BlueSky will pan out with how its protocol is designed. Really pumped to see where this goes!
Cheers!
2/2
Hey, Kuba here, nice to be here!
What I do: Software Engineering Team Lead @ Spacelift and author of OctoSQL
Interested in a wide variety of software engineering topics, esp. around developer-/devops-/data-tooling, but recently also playing with AI trying to make it do something useful.
1/2