chwzr

m first run got 7.83

> end-to-end evals keep the harness honest sorry, but the README shouts: i am slop. no matter how good the code is. what is a honest harness? i would prefer at least a readme written by hand, im okay with ai generated code, but the readme is for humans not agents

so if I understand you correctly crdt is not used to collaborate in realtime with other human beings? instead we need to sync via git?

the design looks neat. + for hexagonal pattern on the backend - that fits perfectly!

I would really love to see an iOS remote control app for zed. I am using it on throw away microvms via ssh. Would love to have the zed server running there, control agents via phone and occasionally use my Mac to connect to the server and use the desktop app as normal for review and hand coding

I see the alpine and python:3.12-alpine images in your cli docs. Where does these come from?is it from a docker like registry or are these built in? Can I create my own images? Or this this purely done with the smolfile? Is there a Ubuntu image available?

Looks really nice btw. Hot resize mem/cpu would be nice. This could become a nice tech for a one-backend-per-customer infra orchestrator then.

Wonder what you are thinking of then?

there is TabPFN [1] which also has time series capabilities.

[1] https://priorlabs.ai/tabpfn

i have this in my .zshrc which provides same functionality:

  lk() {
    if [ $# -eq 0 ]; then
        local output=$(sudo lsof -iTCP -sTCP:LISTEN -n -P)
    elif [ $# -eq 1 ]; then
        local output=$(sudo lsof -iTCP -sTCP:LISTEN -n -P | grep -i --color=always $1)
    else
        echo "find and kill processes listening on ports. Usage: lk [pattern]"
        return 1
    fi

    if [ -z "$output" ]; then
        echo "No listening processes found."
        return 0
    fi

    # Show header + results
    echo "$(sudo lsof -iTCP -sTCP:LISTEN -n -P | head -1)"
    echo "$output"
    echo ""

    # Extract unique PIDs (skip the header row if no grep was applied)
    local pids=($(echo "$output" | awk '{print $2}' | grep -E '^[0-9]+$' | sort -u))

    if [ ${#pids[@]} -eq 0 ]; then
        echo "No PIDs found."
        return 0
    fi

    echo "PIDs to kill: ${pids[*]}"
    echo -n "Kill these ${#pids[@]} process(es)? [y/N] "
    read -r confirm

    if [[ "$confirm" =~ ^[Yy]$ ]]; then
        for pid in "${pids[@]}"; do
            echo "Killing PID $pid..."
            sudo kill -9 $pid
        done
        echo "Done."
    else
        echo "Aborted."
    fi
  }

Do you do something similar to the modifications codesandbox has done to firecracker, regarding mmap ram? (They have multiple blogposts about it on their blog)

Would love to chat about details there

How does it compare to tabpfn?

Is this compatible with the dx dioxus cli? Would love to have a cache layer for it

Sounds interesting, do you have any links to these projects? Could not find multicardz when searching

Oh and there is flok[1] which combines the strudel repl with visuals from hydra. Also there are sclang and other algorave environments available. Everything is synced (with crdts i guess) so it’s live collaborative. Which is nice to remotely jam with friends

[1] https://flok.cc

How is the performance of the mounted host directories? It’s some file server over vsock?

Would love to have the ability to spin up firecracker micro vms in proxmox as an alternative to qemu

Tried to create a canvas like app view, your app froze after some prompts (guess the swift interpreter?) reopening the app and selecting the project again does not help

Is it possible to develop custom oauth/openid plugins?

Would love to hear more of that azure marketplace. Does it bring leads on its own?

I currently tested the cursor IDE [1]. It’s vscode with nice ai integrations made with great ux in mind. They integrate with ChatGPT and Claude Sonnet/ Opus. For my experience (rust/typescript) the sonnet 3.5 model is giving me better results.

You can pass the current file, or multiple files or your whole project as context to the models.

[1] https://cursor.com

- Brightintosh (use the increased brightness (1000 nits) of the integrated XDR display)

- raycast (window management, shortcuts, etc.)

- tailscale (private mesh vpn)

- linear (issue management)

same here, wondering what's going one. Do they release post mortems?

You can rescale cloud servers at hetzner, but you will need to shutdown the server in the process

> This is meant to be installed by a professional contractor, not the homeowner.

But I would want to install that myself. I do everything in my house myself. Could I order a device from you and at the same time save the compensation that otherwise gets the installer?

+1 for react-flow it’s a joy to develop with

Next.js with Prisma ORM - Dream team, you get SSR, Typescript Typings ootb.

If you host on Vercel you need a Postgress somewhere (aws, neon, Heroku, fly.io…)

If you host on own vps: use nosql and litestream for backup to s3(backplate b2) . Nosql is blazing fast on ssd vps

This would be the killer 2050 alien feature. Git is a really good mental model on itself. Git + Neon = you invent the future!

Is there a timeline on releasing orchestration things?

Thx will have a look into dynamobim