I bought a pretty powerful desktop computer for gaming in late 2025. It came with an RTX 5080, which I started using to run some local LLMs and run some experiments (most recently, I was trying to get agents to get better at playing Zork I).
I've mostly enjoyed having WSL to leverage Linux dev tools, but it seems like it's still adding overhead that prevents me from taking advantage of the GPU in full, so I'll likely get another drive and install Linux.
I tried Qwen, Llama, Mistral and Gemma. Gemma 4 was pretty impressive.
I can see how this is a great building block for what Zed is doing around collaboration.
One thing I would really love to see, however, is a way to better attach code review comments to the specific version of code they were left on. I find it quite difficult to do with git and github, considering that commit hashes change every time one is forced to rebase (say, for example, to handle a merge conflict).
Do you expect DeltaDB to help address this problem?
Some programming is the equivalent to designing a factory, some programming is the equivalent to carving out wood pieces.
I've worked in large organizations where a small number of staff+ engineers were approaching programming as factory, while many more engineers were obsessing about the details of their own narrowly-focused work and liked it that way.
It feels like the scale is tipping more towards the factory model and, understandably, not everyone is excited about that.
To be fair, shipping these guidelines as AGENTS.md/CLAUDE.md in the repo that contains the assignments will make it so that agents will pick this up without needing students to opt in explicitly. Seems like a reasonable first step to me
I totally fell for the "obscene memory" trap myself. My first encounter with suffix trees outside of a textbook was for an ITA Software 'Instant Search' puzzle. The requirement was sub-0.1ms search on a large string database, I went straight for a generalized suffix tree. Then I realized they had asked for the solution to fit within a 1GB heap. :(
Author here. It’s a very hopeful time for interactive learning. Coincidentally, it is inspiring that both OpenAI and Anthropic released these improved visualization capabilities just this week.
Using Gemini's canvas for chord progressions is a great example of this. When I was building this suffix tree visualizer, I kept thinking about how much "spatial" intuition is required to understand the algorithm; having these live, interactive environments available to students is a massive step forward.
Hi HN,
I wrote this post after an experiment over the holidays. I had Claude write a cross-stack feature touching our database, cloud infra, mobile app, and the embedded application running on our hardware devices at work. What would usually take me a week took an afternoon to generate.
But it still took weeks to test and merge.
The takeaway for me was that for teams operating with legacy debt, or teams where verification requires physical interaction (you can't throw prompt engineering at a hardware test bench), AI doesn't solve the bottleneck; it just shifts it. We are making code generation incredibly cheap, but the cost of verification and code review isn't shrinking, and the burden is falling on our most senior engineers.
I’d be curious to hear how others managing complex or non-standard codebases are adapting their CI/CD and review processes to keep up with the volume of AI-generated code.
I was asking because if you optimize for pitches response rate, you end up with better pitches but not necessarily good matches between candidates/companies.
I tend to agree with this point of view: whiteboard interviews fail to predict how you will write real code once you are hired, your actual code is definitely a better indicator.
Asking candidates their github usernames is starting to seem a common trend, in my experience.
Sounds like a smart idea, to me. Since I started using Alfred on my Mac I'm able to control almost everything from the keyword and I'm faster at almost everything.
Being able to guess the user intentions from the context would be a great advantage, but the challenge would be balancing context-dependent commands from the most general ones, in my opinion.
I'd be curious to know how much of this applies to non-native English speakers.
I guess their native language can shape the way they build sentences and influence the way they communicate in different way. Or, maybe, they just attempt to correlate more with native English speakers.
I guess that one of the advantages of such an approach is standardization, in particular if your company is big.
Once you get a significant number of candidates, it may become more important to have some "objective" and quantitative way to rank your candidates by performance than to get a true understanding about how likely they are to get things done (how would you "measure" that?).
At that point, it becomes more important to ensure uniformity of judgement across the pool of interviewers. Isn't that why they often ask programming puzzles: so that they can compare performance.
If the price to pay for that is that some valuable candidate won't get an offer, they can live with that.
At least, that's how I see it.
Of course the argument for startups will be different :)
I used Tower for a while, and I didn't have the chance to test Git for Mac yet (will do later today when I'll be using my Mac).
One thing that Tower can do quite well is dealing with multiple remote repositories (e.g. github + heroku).
If GitHub for Mac does that as well, I believe I'm going to switch to that.
I've mostly enjoyed having WSL to leverage Linux dev tools, but it seems like it's still adding overhead that prevents me from taking advantage of the GPU in full, so I'll likely get another drive and install Linux.
I tried Qwen, Llama, Mistral and Gemma. Gemma 4 was pretty impressive.