Realme 5

[...] I've already bought a USB-Floppy (and other USB-based "old media" bits), but I'll definitely be keeping the LS-120 now as I think IDE is more durable (and IIRC LS-120 drives were more "precise" w.r.t. [...]

[...] They demand a reproducible build, but then can't do things like install the right version of yarn (no - npm install -g yarn is not correct, our readme says it in bold like 5 times and provides the exact correct command to install the right version), or follow basic setup steps like "Use this version of node (complete with exact steps to install it and a script to automate that for them)". [...] They're incompetent at best, and I'm fairly convinced they're just milking the google search deal income for as a much as it's worth - I don't think they really want to provide an alternative and secure browser anymore.

I currently use DeepSeek-V4.1-Flash and the V4-Pro and -Flash versions before that, because they are extremely cheap. [...] GLM-5.3, GLM-5.3-Flash and Kimi K3 are also fine, but slower, more expensive, and less good for what I use them for, which is mostly Python and CUDA programming with some JS and HTML inbetween. [...] I haven't tried Astra or Fable yet, because I am not made of money and am happy with my current setup. [...]

[...] Again, if it is an one-in-a-while job, the time taken to build jigs is not important. [...] Having done so much work, I just can't see dropping that kind of money for something that isn't likely to give me any real advantages. [...] Best of all, if you need five people working on a project, making extra jibs and fixtures is easy and cheap.

Really cool, but the thing I really appreciate is that you had this towards the top of the GH README: "Built with Codex using GPT-5.4 (Low)." I am fine with using AI to code something, but I really like it when people are honest about it so that I know what I am looking at.

The biggest day-to-day merge conflict annoyance I have is situations like this. As far as I know, there’s no solution: commit aaaaaaaa diff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -1,4 +1,4 @@ -<p align="center" +<p align="left" <img width="200" src="logo.svg" </p commit bbbbbbbb diff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -1,5 +1,5 @@ <p align="center" - <img width="200" src="logo.svg" + <img width="345" src="logo.svg" </p Project Commit aaaaaaaa changes some text on line 1. Commit bbbbbbbb changes some unrelated text on line 2. I don’t want to have to manually resolve this—just merge the lines. If there’s a semantic conflict I’ll let the tests sort it out. When I’ve looked in the past there wasn’t a merge strategy that fixes this.

[...] I've known this for the last 15 years. [...]

[...] Same with AI, sure we're generating endless piles of code, but how much of it is actually leading to better software?

Why should 6 symptoms constitute ADHD while 5 don't? The person with only 5 doesn't have it? This is not science. Change the diagnostic criteria to a blood test or brain scan and then we are in the realm of proper science. But until then the burden of proof should be on the ADHD community not on those trying to dispute it. It's like having to use proper science to disprove something that is not conclusively proved in science. It's fine to publish all this science and evidence, but it all comes back to the diagnostic criteria which is so wishy washy. The diagnostic criteria has evolved from a completely arbitrary definition. There is no reason the DSM stuff should be used at all, and trying to find science to support a diagnostic criteria that was arbitrary to begin with is silly. That's why, after all this, the main point I make is that the boundary between ADHD and non-ADHD is arbitrary and will always be and its more about having a disorder to point to to be able to prescribe stimulants to "people it would help". Mark my words: there will never be a biomarker or test used to diagnose this condition because it would only reduce the number of people who can access treatment. It's an elaborate exercise on how to be allowed to use a scheduled medication as treatment. If the government didn't ban drugs then all this ADHD diagnosis stuff wouldn't exist at all. It would be the reverse exercise: which people should not be allowed it because they abuse it and it negatively impacts them.

I'm also skeptical partially because I don't like the huge essays generated by any llm. CLAUDE.md/AGENTS.md/README.md that are 5+ pages long are all equally bad imo. I prefer following the idea that if something is too verbose for me to want to get anything useful out of it, then the llm should behave similarly. Even if it's not true, why waste 2 paragraphs explaining something that could be explained in one short sentence? My CLAUDE.md or AGENTS.md is usually just a bulleted list of reminders with high level information. If the agent needs more steering, I add more reminders. I try not to give it too broad of a task without prior planning or it'll just go off the rails. Something I haven't really experimented with is having claude generate ADRs [1] like your RFC/ idea. I'll probably try that and see how it goes. [1]:

I’m a big proponent of writing simple, understandable code, so I’m playing devil’s advocate here a bit, but: who cares? A significant (majority?) portion of developers have been shipping JavaScript/node applications for the last decade that contain hundreds of MB to GB of code from god knows where doing god knows what with dependency trees the size of redwoods. It’s not like your average mediocre dev really knew what was going on behind their gluing of frameworks together - at least from what I’ve seen. If you have remotely competent tech leadership that enforces relatively intelligent patterns (a good one I’ve found is “write everything backend in rust”) you can make AI churn out monstrous amounts of code that… isn’t all that bad? And if you enforce it writing and updating a docs/API.md on every commit/PR you’re probably doing better than 80+% of devs I’ve ever met. Up until a few years ago it wasn’t uncommon to roll up to a new job that was a “legacy” pile of garbage concocted over 20+ years with no comments or API docs and a readme that tells you to ask for help from someone who has been dead for 5 years. At least AI code is full of comments (some of which might even be accurate) and there’s a finite (relatively low!) cost to figuring out “wtf is this doing and how is it doing it”

[...] I didn't really build projects, just scripts. [...]

To say that 5-MeO is a short duration psychedelic doesn't convey what's actually happening -- basically, you go from your normal waking state to likely the most profound experience of your life, and then, if you felt so inclined, back to work...all within 10-20 minutes. That's precisely what I didn't like about it, the rapid shift back to everyday consciousness. At least with other psychedelics you can spend an entire day in the woods, feeling the entire ebb and flow of the experience, and perhaps learn something in that process. With 5-MeO it's just instant God realm, and then, hmmm, what's for breakfast? Strange substance :) Finally, it's not for the feint of heart; until you try it you have no idea what you're getting into -- there is no preparation for the rocketship that is 5-MeO.

All instructions to LLMs are merely suggestions to nudge it in the right behaviour. Unless you have a deterministic guardrail that guards against a single specific action, everything else is a slot machine that's biased strongly in your favour. My global CLAUDE.md explicitly states "When commenting on code and configs, or writing MD files, strictly write within the domain of the content being commented on. DO NOT include information, negatives or ramblings from work sessions. For e.g. if commenting on a proto string field that is replacing an int field, do not comment that 'this is not an int field'". This reduced the idiocy of the agent (Opus 5 included) when writing documents. But I'm still catching it writing README.md talking about the negatives that it removed. Those belong in the memory if it is actually that important (most of the time it's junk), but Claude doesn't seem to understand and never ever learns.

Yeah, speed is the one, I believe the default TypeSafe API quota is 1.5-2k queries per second (batched in bigger requests). On the readme I'm so sorry to tell you that, but it's 100% written by Opus 5.5 with zero "pretty please don't write slop" prompting, it's just how slop is going to look like from now on. I've been writing code for 15 years or sth like that and the code is also what I'd call pretty reasonable..

Owning an EV doesn't necessitate spending a few grand to have a fast charger installed. They charge fine from a regular 110v US wall outlet -- they just do it slowly. And by "slowly," I mean: Somewhere in the realm of 2-5 miles per hour. [...]

We work hard (and continue to invest) to make sure that providers serving sub-par inference can't game the system, and that our routing actively avoids them. I'm glad to hear that y'all are doing this, as I was unaware that this was something OpenRouter does. I was surprised and disappointed that there are so many problematic providers that it seems like community best practice [1] is to ban somewhere in the realm of 5-6 providers. Would it be possible to provide some way to express an even stronger preference for high quality providers? E.g. "only route to first party for this model" or, "cost, but don't route to providers that more than x% worse than the first party". I'm sure something like that can be done via the API but I haven't found a UI way to do it - and having it in the UI would go a long way towards feeling like OpenRouter is looking out for me/helping solve the problem as opposed to leaving it to me to have to figure out. [1]