rafaelmn

6 ms·

91.
by rafaelmn 7mo ago
Lol if CC is the advantage that's the larges indictment of AI coding there is. Don't get me wrong CC gives me good results, but I very much doubt their tooling is great, they just spew tokens at the model and the model is quite go
92.
by rafaelmn 7mo ago
Not tmux related at all had it happen in all kinds of setups (alacritty/linux, vscode terminal macos)
93.
by rafaelmn 7mo ago
I mean if you want glitchy garbage that works in the happy path mostly then game engine is the right foundation to build on. Software quality is the last thing game devs are known for. The whole industry is about building clever hacks to ge
94.
by rafaelmn 7mo ago
It's shocking how shitty claude code CLI app is - config is brittle shit (setting up a plugin LSP is searching through GitHub issues and guessing which parameters you messed up), hooks render errors in the app when there are none and t
95.
by rafaelmn 7mo ago
Your battery is going to suffer because of the extra ram as well. I don't know your workloads, but for me personally 64 GB is the ceiling buffer on RAM - I can run entire k8s cluster locally with that and the M5 Pro with top cores is s
96.
by rafaelmn 7mo ago
It's not even features - basic stuff like input handling/focus is broken on iOS PWAs it's an obviously ignored tech.
97.
by rafaelmn 7mo ago
No not really - because it doesn't compare to other comparable drivers. You're assuming that human drivers can't drive defensively. Like a realistic comparison would be comparing it to taxi drivers/ride share drivers - t
98.
by rafaelmn 7mo ago
If you're arguing to the regulator that "look these cars are not as bad as humans overall" I guess that's fine, but if you're trying to sell me on using self-driving by comparing to something that's completely
99.
by rafaelmn 7mo ago
13x compared to what ? The average driver is such a bullshit statistic - accidents are highly correlated with stuff like alcohol/drugs/lack of sleep/lack of experience/physical issues, then the other huge behavioral fact
100.
by rafaelmn 7mo ago
Honestly if I'm just following what a single LLM is doing I'm arguably slower than doing it myself so I'd say that approach isn't very useful for me. I prefer to review plan (this is more to flush out my assumptions abou
101.
by rafaelmn 7mo ago
> That’s normal and natural and the parents move onto a new phase too. Is it really ? I would say the "natural" way of things is older generation gets supported by children and they help take care of grandchildren while their c
102.
by rafaelmn 7mo ago
I've seen a few instances of where Claude showed me a better way to do something and many many more instances of where it fails miserably. Super simple problem : I had a ZMK keyboard layout definition I wanted it to convert it to QMK f
103.
by rafaelmn 7mo ago
> The "no LLM" stance makes sense if you don't have guardrails. With the right constraints, AI-assisted code is faster AND safer than solo human development — because the agent never gets tired, never rushes before a deadl
104.
by rafaelmn 7mo ago
No my logic is "if everyone had the chance to do what they wanted nobody would do what is needed".
105.
by rafaelmn 7mo ago
The only reason you can "afford" anything is because other people do shit they would rather not, but do so for the money. If people could actually "afford to make choices" you'd find out pretty soon how dependent yo
106.
by rafaelmn 7mo ago
This is basically the performance of M1 with 8GB ram (with shittier USB/connectivity). I've seen developers who used the 8GB air a few years ago on a project. It would't work for me (even the 24GB air I have is swapping), b
107.
by rafaelmn 7mo ago
I don't get where this class/status/worthiness ties into HN comments ? I get decent feedback most of the time, and I read interesting stuff, it's the easiest way I found to stay in the loop in our industry. What are you
108.
by rafaelmn 7mo ago
These people play around with shit and try to sell you on their secret sauce. If it actually works it will come to claude code - so you can consider them practical SOTA and honestly just plopping CC to a mid sized codebase is a pretty great
109.
by rafaelmn 7mo ago
The study you quoted is sonnet 3.5/3.7 era. You could see the promise with those models but the agentic/task performance of Opus 4.5/4.6 makes a huge difference - the models are pretty amazing at building context from a mid s
110.
by rafaelmn 7mo ago
Eventually they'll get rid of juniors and mid level devs because realistically it's easier to review when you're the one doing the prompting.
111.
by rafaelmn 7mo ago
I think I've heard multiple time that a large % of training compute for SoTA models is inference to generate training tokens, this is bound to happen with RL training
112.
by rafaelmn 7mo ago
I don't like that analogy. If I had to work with a Claude like junior I would ask for them to get removed from my team - inability to learn stuff, completely unexpected/unrelatable faliure modes and performance. On the other hand
113.
by rafaelmn 7mo ago
No it's more like if you knew how to build it before - LLM agents help you build it faster. There's really no useful analogy I can think of, but it fits my current role perfectly because my work is constantly interrupted by prod s
114.
by rafaelmn 7mo ago
That's not US prices, it's just price without VAT vs with VAT included. US also has sales tax it's just not included in list prices.
115.
by rafaelmn 7mo ago
I'd say half of that difference is that we have VAT included in price. But my point is that's a 16GB jump for 200$ not 8GB
116.
by rafaelmn 7mo ago
I was configuring my M5 MBP preorder and 48=>64 was 250 EUR so not sure if they cut prices or your numbers are outdated ?
117.
by rafaelmn 7mo ago
Is this working age population or all ages ?
118.
by rafaelmn 7mo ago
Why ? If your device is a thin client for web services/gaming the risk of bitflips/bad ram is a minor annoyance.
119.
by rafaelmn 7mo ago
But those tools aren't really compute bound anyway - you're not buying a workstation to do them, you're getting a consumer laptop or a tablet.
120.
by rafaelmn 7mo ago
> Very few applications scale with cores You mean like compilers and test suites ? Very few professional workloads don't parallelize well these days.

More ›