dayjah

* current mathematicians

Were early in this cycle, we will learn to do more, and exercise our new capabilities more fluently, which in turn will create more skilled practitioners

Consider the abacus, calculator, computer, etc, each of these enhanced mathematicians’ capabilities and thus outputs.

WIF works far better when you don’t want humans in the loop. For example, we’d do our development on cloud instances, those have identity linked to our humans via our IdP. Our IdP governs all access, for example: it lets devs use Datadog. If an agentic workflow needs Datadog access and the MCP requests OAuth that slows the loop down. At the same time, we don’t want Service Accounts everywhere because we need to be able to answer “who” a lot for compliance reasons.

I can’t speak to what they’re actually doing; but there is something here that I encountered in a past job.

Premise goes: our brains are infinitely more efficient at reasoning than current computational architectures. Presumably because our synapse permits allow for non linear connections in a dense space (grey matter). At a biotech I worked at the phrase that described this difference was “volumetric computation” vs “linear computation”; I can’t find a reasonable description of this online so I’m going to assume it was a house term.

This site seems to suggest the internal geometry of the connective/reasoning layer is more important that having more data. This causes me to think “graph” vs “array”.

Or yeah, maybe it’s just slop and some number of people will be creative and buy the hype ^_^

I’ve near infinite respect for DJB, so I’m assuming I’ve missed something obvious here. Why is ‘,’ being used as a termination byte. Is it just a backstop? If buf[len+1] != ‘,’ then there’s a line error?

I’ve done plenty of wire protocol work, and length prefixed strings are great to work with. I’ve also been lucky that those strings were typically contained within a broader payload. To that end, I’ve not had to think about the case of many strings one after another.

I think homogeneity is an unavoidable end game for the internet (unfortunately).

At work we’ve been discussing whether to migrate off our home grown component library to Material UI. I shudder at the thought, personally. However, a compelling reason to use a ubiquitous framework is that the ubiquity means folks intuitively know how to interact with your product.

Like many of us I was born into a deeply customizable Internet, all of my websites were green or red on black. They were a glorious amalgam of fixed width fonts and <blink> tags. With occasional wingdings characters for fun and games and complex <table>/<tr>/<td> tags for really epic layouts. They were l33t, honestly ^_^

But, as time goes on and more and more people use this thing, converging on the one-true-UX feels like a net good thing assuming the fundamentals are right. To some degree the LLM-ization of the Internet is essentially the end game of squashing the personality out of the Internet which bootstrap started.

We’re on the cusp of spoken word being the core UX of computers with a fall back to reading the LLM transcript, neither of which benefits from <blink>

$VC disagrees w/ you on this.

Unfortunately.

Personally, I still pay Zed because I think they’re doing great foundational work. However I’ve now almost entirely migrated to helix for editing needs.

I was feeling this pain also; so I switched my workflow to watching file changes with lazygit, and then switching to helix to make small tweaks.

Another option you may want to try is mux (github.com/coder/mux). It wraps the LLM in a nice interface which has the ability to do line/block comments on changes by the LLM that then goes goes into your next prompt. It’s very early stage though: v0.19.0.

I suppose that would depend on the restaurant / patron.

For example, in Texas there are loads of TexMex restaurants and Hispanic cultures actually embrace children as part of the environment vs Western European cultures (which I was raised in) which don’t so much.

As I said: horses for courses.

Horses for courses. Our we let our 2yo run riot at restaurants while we enjoy our food. It was an adaptation for my wife, for sure. I love it and best I can tell other patrons and staff love how comfortable our daughter is in the environment.

The Attention algo does that, it has a recency bias. Your observation is not necessarily indicative of Claude not loading CLAUDE.md.

I think you may be observing context rot? How many back and forths are you into when you notice this?

Pretty sure I just ran a nix flake update.

Here are the links from my journal:

This went into nixpkgs: https://github.com/NixOS/nixpkgs/pull/376988

Which then changed the api between and broke this: https://github.com/LnL7/nix-darwin/blob/master/modules/nix/n...

The fix took a few hours, I happened to be one of the first folks bit by it: https://github.com/LnL7/nix-darwin/pull/1318

I also have in my notes that Emilazy is a super star: https://github.com/emilazy

Notes on how I worked around it for the time it was broken:

> To work around it on myside I tried various things. Fundamentally I rolled back to nixpkgs-24.11-darwin which needed corresponding changes to nix-darwin (nix-darwin-24.11) and home-manager (release-24.11) to get everything working.

That’s great to know! There are plenty of issues which are reasonably well documented in the zed repo.

https://github.com/zed-industries/zed/issues/26277

About 4mos ago I moved to using brew for zed because at the time there was some hard block on updating rustc in nixpkgs-stable to a version which included some feature that zed relied upon.

This is all great feedback, thanks!

I got here through devenv, I was fully bought in on its proposal and once I found its edges I started peeking under the covers to understand how it worked.

At that point I was pretty deep in mise for everything that wasn’t using devenv. This perhaps help frame why I see them solving the same problem.

I definitely had my “aha!” and ditched mise because nix seemed it had solved my problems. But now, in a new gig, I’m running into lots of edge cases that mise could solve at the drop of a hat and nix (/ my poor understanding of the fundamentals) struggles with.

So, with that all said, I suppose my point is that you get a lot of overlap between the two, and mise is easier to use and get buy-in on. There are certainly elements I find appealing about nix which mise doesn’t touch (promise of repeatable builds, the entire package ecosystem, etc), however.

My principle of adoption was essentially this but in reverse; use it on the system I use the most (macOS), learn, and then use my niche knowledge to apply it to less frequently used computers like my gaming rig.

Along the way I acquired enough talent that use at work seemed reasonable.

As time has gone on, however, I have found things like the stringent need for everything to be built results in archaic packages versions in nixpkgs, etc., while core waits to bump the rustc version. Thus my return to using brew for almost everything albeit managed via nix-homebrew.

Case in point: I use zed, which relies on cutting edge rust features, which nix cannot deploy because of stability concerns. Everyone is right in this situation, but that left me with an archaic version of zed until I moved to the homebrew version.

In practice nix-darwin relies on being a drop in, which means maintaining compatibility with api surface which in the proper nixpkgs world is a closed loop. There are several cases of this breaking since 2020 or so.

I’m about 18mos into managing my macOS hardware with Nix. And I’m conflicted. It’s clearly a powerful system, and I’m still very noob at it. It’s not clear to me that it’s the right solution for macOS. I’ve not felt comfortable enough with it to roll it to Linux hosts yet. Or use its docker image maker.

Consistently through the 25.05 period nix-darwin and nixpkgs would fall out of sync. I learned not to `nix flake update` too often as a result. It’s amazing that rolling back is as easy as it is, and that’s huge, but if you squint and reason that mise and nix solve the same issue, why not use the less opinionated, easier to reason about mise?

As time has gone on, more and more of my system is managed via nix-homebrew … effectively producing a Brewfile for the vast majority of my package needs. Why not just use Brewfile directly?

I really want to advocate for nix, but it feels like I lose the “why not x?” conversations with myself, I can’t fathom winning them against a less invested peer.

Serious question: how in theory?

I’m under the impression you need to radiate through matter (air, water, physical materials, etc).

Is my understanding of the theory just wrong?

Yeah, to that point I’ve been interested in trying out a Shadow Systems CRp. But haven’t found a place carrying them..

Same, fortunately my local shooting range has a good selection of firearms to try. I tried the P320 and P365. I had been considering the P365, but it was so snappy I reasoned the otherwise completely boring P320 would be a better fit for my needs. However, it was close enough to another firearm I own that I decided to cool-off and see if I still wanted it in a month... then this all broke.

When the plaintiff is the US Armed Forces, does this still hold? Surely the contract with an arm of the military doesn't have the same protections?

I found the syncing process in chezmoi to be so hard to mentally model.

I’d often change a file, forget that it was backed by the chezmoi store, later find myself trying to reconcile the differences, just so I could commit and share w/ another computer. nix + home-manager and snowfall lib, once over the multi month ramp up, have been such a breath of fresh air in multi system management

Oh man… that “how normal is this thing of mine” feeling writing nix gives me is such a weird characteristic of the ecosystem.

I simultaneously don’t want anyone to see mine while desperately seeking affirmation that mine is not weird ^_^

Happy to hear I’m not alone!

I do find the ironies in political platforms quite beautiful. I also love how they provides endless fodder for largely fruitless internet discussion ^_^

Hey now. America is a broad spectrum of people — some of us are heretical and believe governments have a role in everyday life, some of us believe the opposite.

If you only ever look at the way a system works at a specific point in time you only observe it at that point in time.

America has had multiple attempts at solutions for healthcare over the years, each started with good intent and then waylaid by various causes to produce what we have right now.

A sibling comment mentions political compromise to pass the ACA, as an example of this.

Another example is that HMOs were started with inherent goodness, but got “corrupted” (in my mind) by profit seeking.

To directly answer your question: a core tenet of the Republican tent is minimal government involvement in day to day lives of the citizenry. Ergo, the Swiss system won’t work because it involves a lot of bureaucracy. Republicans link bureaucracy to cost, and feel this is not an appropriate use of tax payers dollars.

The holes in this political doctrine are not part of my answer here fwiw. Please no “but…” comments to that end :)

For counter argument sake, if a highly regarded company says: “we’re going in a hiring spree”, all of that available talent adds 50% to their comp ask.

Broadly speaking though, I think you’re experiencing confirmation bias to some degree. If you only look at companies that are on the struggle bus, then you only see a limited number of levers that management has (RIF, delay CapEx, etc). Other struggling companies that don’t take evasive maneuvers go out of business and we don’t hear the story.

Honestly, I didn’t ever ask. As the SWE on the project I just had to switch a DIO on/off based on the reading from an AIO. I just trusted that the EE and ME folks were making smart decisions around BOM.

It’s not clear to me how the secrets are referenced in storage. Is the expectation that given `--provider onepassword` that one of the entries in 1p would be “BUCKET”?

edit: it’s not covered in the post, but it is on the launch and doc site: https://secretspec.dev/providers/onepassword/

I worked on a scientific instrument a while ago, it had a Peltier heater on it to raise the sample to 60c from whatever residual environment temp was (approx. 20c in a lab). It was pretty amazing to see my old overclocking cooling solution come back around and be used in my professional career some 20+ years later