IMO it's the same problem with Rust. LLMs are trained on vast quantities of code that violate Rust soundness in safe code.
LLMs are prediction engines: you can move the needle (heavily) on the predictions from lazy to correct by construction, but the defaults even on high end models like Opus are always riddled with shortcuts.
LLMs can produce coherent & safe Rust 1.92, LLMs can produce coherent Zig 0.16, but that's dependent on in context learning & instruction following.
I try not to be negative, but this software seems bizarre. It's a Lua meta manager written in Zig, that calls itself cross platform (Linux / MacOS only), and depends on external gnu build tools, not even the native Zig portable retargeting llvm.
LLMs are usually too busy agreeing to push back on the ideas & details like this.
I still have no clue what you're implying that I was implying? Reading too much into someone's words is definitely a thing.
There used to be 4+ major browser engines. Now there's only two, and Google owns almost complete share & following standards? Google prefers their monoculture.
More diversity in browser engines was a good thing & standards were lovely.
Not everything has to be a slug-fest between #1 & #2.
& personally I'm glad there's Grok & Gemini to keep Anthropic & OpenAI on their toes even more than the competitive band of open weights models already do.
The corporate models tend to be defensive and sound like HR has approved every word of the script. Not a fan.
Personally been experimenting in Lean 4. LLMs understand it, can be given simple rules to improve it. Typing is strong, proofs are solid, and it compiles quickly.
On the contrary, for a small rust project, I had to clean out 180gb of cargo nonsense from the last ~3 days worth of compiles on a single, narrowly focused topic branch.
The library situation might be funky, but I'm also learning Lean 4 by hand. The tooling & lsp integration is lovely.
For a around two years there, the dynalink dl-wrx36 was selling between $50 & $85 for comparable/better hardware. Still running a pair with an nss-enabled fork. Still effortlessly maxes out a gigabit fiber line with qosify.
Configured it so the 2.5gbe port connects towards the lan where a cheap wifi 7 AP can broadcast the additional signal if anything feel like it needs it. But practically speaking nothing does.
While the openwrt one was a decent experiment, it was far from the only hardware in the price range that had stellar openwrt support before/after it came out. And one thing people seem to forget about with a lot of options (like banana pi options) is that the range & falloff can be terrible. The openwrt two is apparently delayed & going with a different manufacturer.
Openwrt is great if you are willing to customize the software especially. The fact that it can be used as an actual wifi client in a pinch is also a lifesaver.
Long-long term availability is a different problem, but different manufacturers move on.
It always feels a bit vexing when people complain about skills: Personally, we treat them as if manuals where the goal is to patch knowledge, not (typically) be a from-scratch primer.
Letting an instruction following llm deep research and iterate has given fantastic results before.
Being able to construct non-trivial Zig 0.16 programs without slowing down for version-hallucinating compilation errors is nice as a random example.
Note as of yesterday, they retired the lite coding plan you're talking about. New buy-in is $50/month for the pro plan unless you were already on the lite plan.
I feel like I have to mention "The Sky So Big and Black" by John Barnes. IMO, rather underrated. Hadn't really read any good Mars-based science fiction before.
Haven't dived deep into it yet, but dabbled in similar areas last year (trying to get various bits to reliably "run" in-context).
My immediate thought was to want to apply it to the problem I've been having lately: could it be adapted to soothe the nightmare of bloated llm code environments where the model functionally forgets how to code/follow project guidelines & just wants to complete everything with insecure tutorial style pattern matching?