Live data from Hacker News

Viewing profile — ic4l

ic4l

HN member
Joined
Sat, Mar 04, 2017, 7:04 AM UTC
HN karma
856
Public activity
107 items

About ic4l

No profile information was provided.

Recent public activity

  1. comment
    Comment #47775255

    The Backblaze Restore Application does not work reliably when restoring a large number of files. On an M4 Mac, it repeatedly crashed and ended up in a non-recoverable state, forcin…

  2. comment
    Comment #45641556

    We tried this initially FROM mirror.gcr.io/{image_name} We received failed to resolve source metadata for mirror.gcr.io/ So it looks like these services may not be true mirrors, an…

  3. comment
    Comment #45641329

    This broke our builds since we rely on several public Docker images, and by default, Docker uses docker.io. Thankfully, AWS provides a docker.io mirror for those who can't wait: FR…

  4. comment
    Comment #44372053

    Thousands of large Facebook groups appear to have been taken down, with many admins reporting a wave of rapid content reports from a single account targeting the group's first post…

  5. story
  6. comment
    Comment #43900664

    Yeah, thats one of the best ones I have seen, and it popped up a while ago.

  7. comment
    Comment #43590178

    That was fast :o

  8. comment
    Comment #43200481

    Did they already disable it? When using `gpt-4.5-preview` I am getting: > Invalid URL (POST /v1/chat/completions)

  9. comment
    Comment #43178810

    `Especially in a database that doesn't allow wildcards`

  10. comment
    Comment #43178800

    I agree with you. Knowing the exact column names can speed up an attack and, in some cases, make it more feasible. Why don’t they just request disclosure of what’s actually stored …

  11. comment
    Comment #43114659

    Great answer, thanks.

  12. comment
    Comment #43114108

    LXD seems like an unusual choice when Kubernetes already has cadvisor and strong monitoring integrations. Avoiding extra agents is nice, but does this really scale better than exis…

  13. comment
    Comment #42937397

    For me the o models consistently make more mistakes for me than Claude 3.5 Sonnet.

  14. comment
    Comment #42836218

    Most browsers won’t report anything higher than 8, even if your system has more memory. I think its related to fingerprinting attacks.

  15. comment
    Comment #42615606

    In case you are wondering this actually beats the game: setInterval(() => Array.from(document.querySelectorAll('.main-btn, .upgrade-icon, .press-collect, .press-btn, .collect')).fo…

  16. comment
    Comment #41330778

    HN stripped the query param for x/twitter. https://x.com/search?q=%22%40cursor_ai%22&src=typed_query&f=...

  17. comment
    Comment #41330766

    Account settings not accessible, which means any pro users cannot use pro features.

  18. story
  19. comment
    Comment #41313412

    Yeah I wasted 10 minutes on this, and even contacted support. Their status page is misleading and the only way to see it is to look at the most recent tiny sliver of their chart. T…

  20. comment
    Comment #41313408

    Started around 40 minutes ago. The API appears functional but any playground login, or mobile sessions seem to be invalidated without the ability to log back in. Heres a twitter/re…

  21. story
  22. comment
    Comment #36560297

    How many people even have access to the gpt-4-32k, or gpt-4-32k-0613 models. I think they give everyone access to the gpt-3.5-turbo-16k, but I have not found a way to request acces…

  23. story
  24. story
  25. comment
    Comment #35963159

    Absolutely, Zack. The work you've done with PAKman is seriously impressive. The efficiency you've managed to achieve with Alpine packages and LXD seems to be a real game-changer. I…