manmal

21 ms·

571.
by manmal 1y ago
Where I live, people have issues getting hired past age 50, and it’s near impossible past age 60. It’s a stroke of luck if you score a job then. So yes they can raise retirement age, but that will just create misery for all those who can’t
572.
by manmal 1y ago
If the LLM wrote a harness and proof of concept tests for its leads, then it might increase S/N dramatically. It’s just quite expensive to do all that right now.
573.
by manmal 1y ago
We're quite happy with mermaid though, not sure what's the issue with it?
574.
by manmal 1y ago
We use it a lot for internal documentation, eg in Markdown on GitHub (renders natively), Notion, and Jira.
575.
by manmal 1y ago
That’s a pretty major downside, since broad in-browser support (GitHub, Notion etc) is arguably Mermaid‘s USP.
576.
by manmal 1y ago
My problem with 5. is that there are many unknowns, especially when it comes to agents. They have wildly different system prompts that are optimized on a daily basis. I’ve noticed that Gemini 2.5 Pro seems way dumber when used in the Copilo
577.
by manmal 1y ago
Claude Code as a tool call, from Copilot‘s own agent (agent in an agent) seems to be working well. Peter Steinberger made an MCP that does this: https://github.com/steipete/claude-code-mcp
578.
by manmal 1y ago
I think qwen3:R1 is apples:oranges, if you mean the 32B models. R1 has 20x the parameters and likely roughly as much knowledge about the world. One is a really good general model, while you can run the other one on commodity hardware. Subje
579.
by manmal 1y ago
Technically they don’t, but OpenAI must be injecting the current date and time into the system prompt, and Gemini just does a web search for the time when asked.
580.
by manmal 1y ago
Maybe time to find an employer with a better culture? I rarely have meetings that I would be comfortable skipping.
581.
by manmal 1y ago
Yes there are MCPs for git and Jira. I‘m not sure about the utility with the current context sizes.
582.
by manmal 1y ago
This tradeoff will be great for self hosted LLMs, because they don’t need large scale batching usually, and less great for cloud providers that do.
583.
by manmal 1y ago
They could read the whole git history and have all issue tracker tickets in the context, and maybe even recordings from meetings. It remains to be seen though if such large context will yield usable results.
584.
by manmal 1y ago
Might have been the case, but no longer: https://docs.github.com/en/copilot/managing-copilot/managing...
585.
by manmal 1y ago
I guess the idea is vibe coding while laying in bed or driving? If my kids are any indication of the generation to come, they sure love audio over reading.
586.
by manmal 1y ago
That reminds me of https://everyuuid.com/
587.
by manmal 1y ago
Fortunately, there is Bluey. I think this show will have a more positive impact on our society, long term, than any other show.
588.
by manmal 1y ago
Kind of, it’s a declarative framework like React & co. Under the hood it maps to either UIKit components or GPU (Metal) rendering. And view identity is very important for change detection. AFAICT, putting a branch in a ForEach invalidat
589.
by manmal 1y ago
In the presentation, they highlighted that the changes it makes are minimally invasive. That sounded as if it’s less prone to going on vision quests, like Claude often does. Is that something you‘ve observed as well?
590.
by manmal 1y ago
It’s a bit niche for HN, but SwiftUI rendering works way better when following this. In a ForEach, you really shouldn’t have any branching, or you‘ll pay quite catastrophic performance penalties. I found out the hard way when rendering a ma
591.
by manmal 1y ago
Depends on the LLM, recent Gemini models are quite good in this regard.
592.
by manmal 1y ago
What kind of guardrails did you give the agent? Like following SOLID, linting, 100% code coverage, templates, architectural documents before implementing, architectural rules, DRY cleanup cycles, code review guidelines (incl strict rules ar
593.
by manmal 1y ago
Thank you, that’s really nice actually!
594.
by manmal 1y ago
o3 is shockingly good actually. I can’t use it often due to rate limiting, so I save it for the odd occasion. Today I asked it how I could integrate a tree of Swift binary packages within an SDK, and detect internal version clashes, and it
595.
by manmal 1y ago
Having watched Slow Horses recently, I immediately recognized the building. My employer‘s HQ is near Barbican too, such an underrated part of the city.
596.
by manmal 1y ago
This video explains the direction-encodes-trait topic very well IMO: https://youtu.be/wjZofJX0v4M It’s the first in a series of three that I can very highly recommend. > there's no way that there's a 1-to-1 cor
597.
by manmal 1y ago
Don’t the high end embedding services use a transformer with attention to compute embeddings? If so, I thought that would indeed capture the semantic meaning quite well, including the trait-is-described-by-direction-vector.
598.
by manmal 1y ago
I’ve been down the observability rabbit hole recently, and what you’re describing is probably a mix of head and tail sampling: https://docs.honeycomb.io/manage-data-volume/sample/
599.
by manmal 1y ago
I think community is a (the?) source of joy, and it‘s unfortunate you‘re having these issues. And I wouldn’t just blame it on being picky, there could be root causes for this that are not obvious - like burnout, other mental issues, brain h
600.
by manmal 1y ago
I’ve seen people do this with Claude Code to great success (not in a GH Action). Even multiple sessions concurrently. Token budget is the limit, obviously.

More ›