Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
anygivnthursday
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
5 ms
·
31.
▲
by
anygivnthursday
4mo ago
Both are bad, what you describe is very real, but so is the opposite. That one fire in one place can end up in a total rewrite of numerous layers because the abstraction never anticipated certain things to happen.
32.
▲
by
anygivnthursday
4mo ago
On most platforms you can usually opt out of targeted/personalized ads that require tracking or collecting personal data, if you do you still get general ads, but that isnt subject to GDPR if they dont use your data for it, I guess.
33.
▲
by
anygivnthursday
4mo ago
And there are also papers that tell thats effective https://arxiv.org/pdf/2510.04950
34.
▲
by
anygivnthursday
4mo ago
No, thats not what gets automated. My comment was for the parent where IC roles transformed to agent managment roles, which I think just another role for agents to manage - themselves. In other words coordination of other agents, not humans
35.
▲
by
anygivnthursday
4mo ago
And isnt that management job also tomorrow's disappearing IC role? A management agent seems quite possible to me.
36.
▲
by
anygivnthursday
4mo ago
Thats my thought as well, LLM agents put you in the role of a (often micromanaging) tech/team lead of a small team, but the speed and fast feedback loop makes it look different.
37.
▲
by
anygivnthursday
4mo ago
LLMs can now quite trivially write Terraform or Ansible code to setup infra, maybe harder for non-engineers to make those prompts for now. But ongoing maintenance and monitoring would require an online agent evaluating signals and deciding
38.
▲
by
anygivnthursday
4mo ago
I asked a question about an openssl s_client parameter and warned me that I need to talk to Opus about cybersecurity lol. FWIW I dont see much improvement and still see quite the same old annoyances, so far I would not pay extra for this fo
39.
▲
by
anygivnthursday
4mo ago
Yeah, it makes the same old errors, being confidently wrong then sorry... I mean, it is still an LLM
40.
▲
by
anygivnthursday
4mo ago
I read Ghostty runs in a single process, but whenever I tried something like that eg a client/server model in urxvt or foot, I ended up reverting, because eventually some weird state affected the daemon and had to restart it killing al
41.
▲
by
anygivnthursday
4mo ago
With todays LLMs, yes. But if they can ever reach a level of a contractor in a reliable way and companies offering them willing to take responsibility (because confidence is high and rest is insured), then one can just hire a cheap AI agent
42.
▲
by
anygivnthursday
4mo ago
Maybe this is what they meant by Internet Explorer mode in Edge
43.
▲
by
anygivnthursday
5mo ago
correct link https://ec.europa.eu/commission/presscorner/detail/en/ip_26_...
44.
▲
by
anygivnthursday
5mo ago
Late edit, I wanted to clarify I do not share my user home, but the VM user home for backups, thats separate user that does not have my own users credentials etc
45.
▲
by
anygivnthursday
5mo ago
I dont think Claude itself needs much, its more like what you do with it. In my case it is doing some gradle builds and java tests with some postgres docker containers inside the vm so I gave it max 8G RAM with 4 cores and have no issues. I
46.
▲
by
anygivnthursday
5mo ago
You mean by default or it cannot be configured that way? I believe, I had Chrome configured to not allow storage by default, only for sites I added to an exclusion list. I cant remember now, but isnt there also an option to change the defau
47.
▲
by
anygivnthursday
5mo ago
This. I am running Claude in its own QEMU VM, it has git access to my project only if I explicitly unlock the ssh key for it. The other day I realized it trying to push a change, it didn't have permission, so it went looking for "
48.
▲
by
anygivnthursday
5mo ago
I regularly see Claude Opus 4.7 dropping constraints from an otherwise small CLAUDE.md at merely 20% context use. I have to keep reminding it, and it has all info ready in its context, still time to time decides to ignore parts.
49.
▲
by
anygivnthursday
5mo ago
Not everyone is in eurozone, there are countries with instant QR payments that are not SEPA. It is nowdays quite common to get a QR with your invoice, online or in person, it might not be just SEPA. But for now cards are still more accepted
50.
▲
by
anygivnthursday
5mo ago
And OPNSense boxes
51.
▲
by
anygivnthursday
5mo ago
Doesn't Slovakia already have its own instant QR payment system? Edit: oh, sorry, just realized you talking about international not domestic payments
52.
▲
by
anygivnthursday
5mo ago
From CI pipelines with pinned package versions and OIDC instead of permanent secrets limited, least privileges.
53.
▲
by
anygivnthursday
5mo ago
I think Google took a different approach with gVisor and limited userspace kernel interface, tradeoffs...
54.
▲
by
anygivnthursday
5mo ago
Congrats! I quit around the same time cca 2012-2014. I did not smoke on a plane, but smoking on trains (and many places indoor) was "normal" before like 2010 around my place. I did not like it even as a smoker and rather went out.
55.
▲
Rob Pike: Tech industry losing its soul
(hachyderm.io)
30 points
by
anygivnthursday
5mo ago
|
1 comments
56.
▲
LLM Hallucinations in the Wild
(arxiv.org)
4 points
by
anygivnthursday
5mo ago
|
1 comments
57.
▲
by
anygivnthursday
5mo ago
> Large language models (LLMs) are known to generate plausible but false information across a wide range of contexts, yet the real-world magnitude and consequences of this hallucination problem remain poorly understood. Here we leverage
58.
▲
by
anygivnthursday
5mo ago
That isn't new, though. Managers often pushed unrealistic timelines and showed lack of care about tech debt well before vibe coding, just the timelines where different, and the magnitude will be bigger this time. But we also have LLMs
59.
▲
by
anygivnthursday
5mo ago
But Mythos is not marketed as a tool that can do the same as other tools already available maybe slightly better, but as a revolution.
60.
▲
by
anygivnthursday
5mo ago
You can bootstrap from your custom ISO with some embedded starter key, upload ISO, loads into RAM and opens SSH, connect and run a playbook to encrypt the root drive where you deploy your OS with your SSH key. If you went with encrypted roo
More ›