udave

Im not sure if transfer learning is still applicable to these models, but if it is then we can just train the last few layers and get the model to ace in a related domain ( going from SWE -> Cyber ).

Hope it survive the next few years of codification of all of humanities unfinished cheap ideas.

Once I tried looking up whether you can configure GitHub’s merge queue to skip MQ checks for PRs that have already completed CI against the latest main commit. I knew from how MQ works internally that this wasn’t possible, but I wanted to confirm. The AI overview on the page said “yes, GitHub has this feature,” and that answer raised my eyebrows. My mind instantly spun up 10 threads thinking through how it could be possible. I typed out how in the AI chat, and the next message said “No, it’s not available". I don't trust AI overview since then, but since its positioning makes it hard to overlook.

lovely website animation :D

lovely collection. Thank you!

Ive made my own agentic IDE centered around worktrees and containerization. it allows me to run multiple development environments on my machine with each development server running in parallel, allowing me to spin up feature branches and test them instantly.

seems like the same trick as behind labelling price as $99.9

try distributing this spec amongst your team members, ask each of them to drive it to completion. no follow up edits. deploy to individual environments and then run a rigorous test suite against all of the deployments. see if all of them behave the same way.

Let's be real, if you and I both ask claude to generate a feature on the same project, what are the chances that it spits out 100% replicated code? But if we are to build the project using a Dockerfile, we will get the same binary and the same image. Products around LLMs are non deterministic unlike compilers.

its a good idea in theory. But capitalists will make sure this does not happen because greed is never ending. today its AI, 10 years down the line it something else. Hence i think the right to be lazy is for a lucky few.

strongly agree. we keep giving away trust to other entities in order to make our jobs easier. trusting maintainers is still better than trusting a clanker but still risky. We need a sandboxed environment where we can build our software without having to worry about these unreliable factors.

On a personal note, I have been developing and talking to a clanker ( runs inside ) to get my day to day work done. I can have multiple instances of my project using worktrees, have them share some common dependencies and monitor all of them in one place. I plan to opensource this framework soon.

wasn't spotify started out as a collection of pirated songs? somethings go in full circle I guess.

that's true for some folks out there. But, ultimately its about these 3 questions: - what you are? - what you want to be? - when you want to be there? I think if you don't have an answer to the last question, you should be fine with 0 efforts.

the sidebar was the best feature in Arc imo. I gave zen a shot just because of that and it was not a great experience to be honest. First, migration was buggy, then the sidebar lacked some basic features like renaming the tabs even though it looked similar. Nook seems to follow in the same footsteps I just hope that they nail the sidebar like Arc. Tab management is a mess and this has so much potential. All the best to both Zen and Nook.

too early at the time of writing, but did mixpanel got breached due to the recent NPM worm `Shad'hulud`? just speculating.

I find the distinction between queue and pub sub system quite poor. A pub sub system is just a persistent queue at its core, the only distinction is you have multiple queues for each subscriber, hence multiple readers. everything else stays the same. Ordering is expected to be strict in both cases. The Durability factor is also baked in both systems. On the question of bounded and unbounded queue: does not message queues also spill to disk in order to prevent OOM scenarios?