the__alchemist

You can use simple heuristics to get a very good idea.

1: How high in the sky is the sun? (Many other factors like time of year, time of day, and latitude feed into this) 2: What is the chance for you at a given time of the sun being blocked by clouds, buildings, etc?

The statement you quoted is very useful [One of the most useful heuristics], and your statement is misleading. You accumulate damage (Photoaging, cancer risk) dramatically faster at high sun angles (i.e. mid day, close to the summer solstice, closer-to-equatorial latitudes) etc than low ones.

I am not trying to be negative, but this is very important if you have pale skin. Being outdoors in morning and late afternoon is dramatically less damaging than mid day.

How does 6-Luna xhigh compare to 6-Sol medium? Or more broadly newer/bigger model with lower effort vs older/smaller higher effort?

Tangent: are "LLM" and "Agent" synonymous?

Nailed it, and have had this response before. Summarized:

  - This is a real problem with JB IDEs
  - It is not specific to an individual's setup
  - It has existed for many years
  - Anyone at JB who actually is interested in diagnosing or fixing it does not need to extract details, support tickets, or logs from users, as it's trivial to reproduce on any system.

I may have observed this regarding performance. (RR + Pycharm) Perhaps the classic software move of: "[My] computers have gotten faster, but JB IDEs are still slow/unresponsive, so they are probably getting slower / more resource intensive".

They will bring a 9950x with nothing else running to a crawl. Then complain they are out of memory. (When it is obviously the IDEs themselves running the system out of memory)

I <3 JB IDEs for their functionality, but resource pigs they are.

LFT the vids! Is there a good way to break down the overlap etc between:?

  - Junie
  - Air
  - AI autocomplete
  - AI Assistant
  - AI Chat

I could def go for some ACP RN.

I don't have a good direct comparison to that, but broadly:

  - Natural treatment of multi-file projects
  - Introspection
  - Refactoring
  - Keyboard shortcuts, specific types of search etc
  - The build/run windows

Overall, it feels like JB treats the core units as languages and multi-file projects, while others, including VsCode, treat it as text files. Whenever I try something else, I look for these features, assume I'm doing something wrong, then come to an understanding that no one else does them to the same degree. For example, Zed has no concept of moving items. I don't know if VsCode does, but whenever I try that, it seems to have only superficial ability to introspect on function calls, data structures etc.

PyCharm and RustRover. It is astonishing no other IDEs have offered direct competition. I would prefer a faster/lighter alternative to my tablet, but none exists.

edit: Sorry; misread. No, and I am very confused by their diaspora of AI offerings!

I wasn't able to integrate Air into my workflow on my first attempt. I'd hoped it would unify Claude and OpenAI, but Claude doesn't let you use it unless you are on the Token plan, which IIRC only makes sense for enterprises. I did not see any immediate advantages in Air over the Codex application, but will try it again in a few months.

I still love RustRover and PyCharm! Switching to anything else would be a big downgrade.

Side note: I am confused by the various JB LLM integrations. Air, Junie, at least 1 (maybe 2 more?) the IDEas have suggested I try.

This sounds similar (The same concept?) to Gell-Mann amnesia; substitute news/media articles for LLMs!

It has not escaped my notice that a significant portion of these are of the form "Design a protein that..."

I learned of the Triune Brain through Carl Sagan's book The Dragons of Eden. It was an engaging, story-telling approach to a somewhat, in hindsight controversial concept. It sounds (oversimplified) like the paper the article describes here is an embryological validation of this model.

No, but I posted on the existing one. Sounds worth trying.

edit: Done

No. Still heavy as hell. (AKA uses loads of ram and CPU, slow, sometimes unresponsive). It is the price we pay - the tradeoff is IMO the unmatched editing and introspection/refactoring capabilities. I'm surprised there hasn't been a real competitor in this space.

Jetbrains (PyCharm and RustRover) for projects; Sublime for one-off files.

Here: https://github.com/zed-industries/zed/issues/38109

I've lost faith in their editor and GPUI - you were the one, Zed! They are full-steam AI on things like this. I expected this article to be about a general alternative to git, but it is about LLM-focused coding in a way that doesn't make sense to me, as it slices a line between human-written and LLM-written code which does not exist in a meaningful way.

Their primary product, the Zed editor, is unusable for me and others due to it mismanaging file syncs on disk - when external edits happen, the editor retains stale state unless you close and re-open the specific file (Not even an editor reopen syncs it). There is a significant risk of your changes being silently overwritten or conflicted. Amusingly, the risk of this is increased when a file is changed asynchronously, which for me, most happens due to a pull or LLM edits!

It begs questions like: "If coding has changed so that we should use an LLM-focused source control tool, why can't the LLM fix a severe bug in our software that has a closed-form solution?"

Holy shit, I don't know how I missed that. Ty! I must have been only looking at the bare parts docs and not the DK.

This one slapped!

Note: I had very good luck with DosBox on a pile of games from that era and older lately. (Edutainment mainly). I didn't try this one though. Now I want to...

I should try that lib! (Same one mentioned in the article?)

I should also give layout/routing a try. My thought was, schematic design, from the perspective of what LLMs are good at, seems like the much more appropriate/easy tasks, so if it's failing at the basics there... I think I also experienced some consternation at patterns I've seen in real PCBs like a section dedicated to bypass caps that have no context of what pins they go to. (e.g. power connected to ground).

And I gave it a board that had a mixed of components/routing that I have used successfully one or many times, and parts that I am unsure about/don't know how to use and wire. Me seeing it make substantial changes to the known-working parts (My stake-in-the-ground), I got very nervous!

KiCad 10

It's an Atmosic Atm3430e, which includes RF energy harvesting. So, I'm using 3 power sources depending on the application: Coin cell, USB, and RF, with a 1v8 net for sensors. Its power pin routing has some particulars that are poorly documented, and there are no publicly available reference designs. There are a few design docs which help, but leave open questions. So, I used LLMs to help me figure out which pins to connect to which. They did not seem capable of actually doing the work directly, but they answered my questions. We will see how correctly in a few weeks...

Also used LLMs to try to figure out how to flash/debug it. The setup I normally use (Tac-Connect ST-Link 6 pin + Probe-rs) won't work here; it's cortex-m, but their official guides indicate they want you do use a Segger J-Link, which I'm not ready to pay for at this point. It is a very expensive thing. So I'm going to try an NXP MCU-Link Pro. The St-Linkv3 I've been using will likely not work.

Hey - I ordered a board yesterday. I used Opus 5, Astra, and Sol at points to QC the board, and offer power routing advice. (The MCU is a brand new one, and poorly documented unless they send you an NDA and the full docs).

Those models were not capable of even schematic editing or building. (I didn't bother with routing or placement). They made gross errors all over the schematic; changed the known-working parts and net routing I'd copy+pasted from previous boards, changed I2C to SPI for no reason, and generally produced an unusable, unreadable result. Overlapping symbols and labels etc.

I will give you an outsider's perspective on an analogy in this case. It is easy to see Candle as a ML crate to use for neural networks in rust. I have used it, and it works well.

The analogy is Tensorflow 5-10 years ago. It is popular, and there are lots of material on it. You quickly learn from talking to people that due to whims, a collection of reasons, people's love of consensus that no one is recommending it; new people are not learning it. In this case, the Torch analogy is the Burn lib.

I'm looking forward to trying these when they stabilize! I currently use WGPU for graphics, and cudarc for CUDA.

Note: Cuda-oxide is similar to Cudarc's host component, but uses a rust-style kernel dialect. Advantage: Share structs between host and device. Disadvantage: Trading standard Cuda kernels for a new, WIP dialect.

I haven't tried the tile API yet; looking forward to it.

The last time I checked, Cuda Oxide was Linux only, and required Async; these are why I haven't tried it yet.

AoE2. Whether you can label that non-competitive is questionable, however.

> You think you do, but you don't.

Nailed it. People want patched versions of the same RTS. AoE4 came out a few years ago as an IMO fantastic RTS which smoothed over many of the idiosyncrasies of AoE2. From a multiplayer perspective, it was doomed from the start, for not being AoE2. I couldn't even convince my friends to play it with me. (Everyone is still playing AoE2, and expansions + patches are still regular)

Great insight. My (much less sophisticated) mental model is: 1: I don't know how to get things to auto-vectorize nor evaluate if they did. [I should learn]; I assume they do not. Manual SIMD it is! [I have a lib for floats and vectors which mimics core::simd but on stable and x86 only)

I will check out your insights regarding matrix-vector ops and C libs!