JulianWasTaken

Correct. Again, I don't think the implication you're making (that someone reading the original comment will think that is safe from a data recovery standpoint) is one any reasonable reader will make.

> Amazon's Matt Wood isn't worried: "If something does happen or we have a power event or there's a flood in one specific location, that data is held redundantly in other locations as well."

> Pogue asked, "I don't mean to give anyone ideas, but let's say I figured out that one of these unmarked buildings was an AWS data center, and I blew it up. Are you saying that it's so backed up and redundant that you probably wouldn't notice?"

Seems pretty silly to argue, but he certainly did say "one specific data center", and I don't think anyone even non-technical will conclude "it's safe if they all go down at once" from this statement.

Quantum Rise | Chicago or NYC (preferred), Remote for the right candidate | Full-time | No visa sponsorship | https://www.quantumrise.com/

Quantum Rise is an AI-forward consulting startup serving the mid-market in the US. We see companies in multiple industries struggling to understand how to leverage artificial intelligence in their day-to-day operations, trying to understand what their existing workforce now needs to know, or trying to understand whether they're about to be disrupted and need to change how their existing processes work. Many have business problems which "traditional" machine learning could help with even prior to the current GenAI hype cycle, but for whatever reason hadn't prioritized doing until now. Many have employees looking to incorporate agentic AI safely and effectively across different business areas. We try and help getting a handle around these issues and designing outcome-driven solutions.

Agentic AI Solutions Manager: You'll work directly with clients to benchmark, train and deploy agentic systems for industry-specific problems. We're looking for someone deeply embedded in agentic engineering practice across vendors and platforms. Companies are looking for ways to leverage agentic systems to upskill their practitioners and need help navigating a new, powerful scary world.

AI Solutions Engineer: You're someone who has experience across the entire spectrum of applied automation, machine learning and agentic engineering.

We think of these two roles as complementary -- agentic AI solutioners help clients understand the forefront, whereas pure AI solutioners help them pick the right solution not the one they saw in the news.

Both are front-facing development roles for someone who loves both tech and tackling real business problems. We love people with prior experience in multiple industries.

To apply, send me an email at julian.berman at quantumrise.com, mentioning HN and including your CV.

(PyPy was not abandoned.)

I assume you might mean to ask "why wasn't PyPy adopted in some formal way into CPython" rather than a separate project, for which the answer is at least partially likely to be because it's a completely separate implementation.

Always hard to tell.

Do you have a source, that seems unlikely at face value to me, though I've never gone and looked for perception studies myself.

In my case (with beancount) an example transaction looks like:

  2025-12-04 * "CON ED OF NY CECONY 251203~ Tran: ACHDW"
    Assets:Schwab:Checking                                                                       -72.33 USD
    Expenses:Utilities:Electric:Supply:Rate                                                      143.00 KWH @@ 18.06 USD ; Supply 143.00 kWh @12.629¢/kWh
    Expenses:Utilities:Electric:Supply:Fees                                                        0.68 USD             ; Supply Merchant Function Charge
    Expenses:Taxes:Other                                                                           0.45 USD             ; Supply GRT & other tax surcharges
    Expenses:Taxes:Sales                                                                           0.86 USD             ; Supply Sales tax @4.5%
    Expenses:Utilities:Electric:Delivery:Service                                                  21.95 USD             ; Delivery Basic service charge
    Expenses:Utilities:Electric:Delivery:Rate                                                    143.00 KWH @@ 25.08 USD ; Delivery 143.00 kWh @17.539¢/kWh
    Expenses:Utilities:Electric:Delivery:Fees:Benefit                                            143.00 KWH @@ 0.71 USD ; Delivery System Benefit Charge @0.497¢/kWh
    Expenses:Taxes:Other                                                                           2.29 USD            ; Delivery GRT & other tax surcharges
    Expenses:Taxes:Sales                                                                           2.25 USD            ; Delivery Sales tax @4.5%

which I extract from the PDF bill I get from them.

Ooh hmm that's an interesting idea. I have wanted a reason to buy an Emporia Vue or something like that.

Quantum Rise | Chicago or NYC (preferred), Remote for the right candidate | Full-time | No visa sponsorship | https://www.quantumrise.com/

Quantum Rise is an AI-forward consulting startup serving the middle market, across multiple industries. Clearly many companies are struggling to understand how to leverage artificial intelligence in their day-to-day operations, to understand what their existing workforce now needs to know, or to understand whether they're about to be disrupted. Many have business problems which "traditional" machine learning could help with even before the current GenAI hype cycle, but for whatever reason hadn't prioritized doing so until now. We try and help getting a handle around these issues and designing outcome-driven solutions.

Agentic AI Solutions Manager: You'll work directly with clients to benchmark, train and deploy agentic systems for industry-specific problems. We're looking for someone deeply curious about agentic systems across vendors and platforms. Companies are looking for ways to leverage agentic systems to upskill their practitioners. This is a front-facing development role for someone who loves both tech and tackling real business problems. To apply, send me an email at julian.berman at quantumrise.com, mentioning HN and including your CV.

Data Engineer: You'll work directly with clients and application developers to design and build data solutions for AI-enabled applications, Lakehouses, Warehouses, and analytics that solve industry-specific problems. We're looking for someone passionate about delivering outcomes that combine best practices for deterministic data management and governance with AI functionality across tools, vendors, and platforms. Our clients need to leverage AI to solve specific problems using data. This is a client-facing development role for someone who loves to work with data technologies and tackle real business issues through genuine innovation. To apply, send John an email at john.swift at quantumrise.com, mentioning DE and including your resume/CV.

I went the "build a few PDF statement parsers" route.

Some I wrote by hand using PyMuPDF, some I coerced Claude into writing (again using PyMuPDF) by uploading a sample bill (I'd never put my own data into an LLM but it's nice being able to find a sample bill, gets it close enough to correct that I can do the remaining bits if there are variations in bills over time).

Overall it's effort (and yes certainly a bunch effort for manually downloading transactions). The financial industry is very behind on this stuff clearly. I'm not sure in a few years whether I'll still think it's worth the effort I put in, which has gone down over the past few months as I automate things, but until it stops being fun I'll keep going.

Nice! That sounds really useful; in my case the KWH usage (and price/KWH) I pull directly out of the ConEd bill, so my only chance to notice those sorts of things would be post hoc looking back in time for big jumps in rate or usage I think.

But good to hear the positive story side for this.

Fun, thanks for letting me know, will remove it :)

(I'll still stick with "I never really have run into a version issue for things I use Homebrew for, for places where it matters, I have whatever-programming-language-lockfile-for-the-project-I-am-developing" for cases where I need to be sure the setup is reproducible, which is why I've clearly never noticed this file was useless).

I have my electricity bills tracking into a KWH commodity. It has been... effectively useless :D.

beancount is definitely fun. I also jumped on this bandwagon in 2025 and it's been a great archaeology experiment of digging through old emails and trying to find as much data as I can about what the heck this random checking transaction is from 2012.

I think a nice thing about beancount is that given how simple it is you can almost even ignore whole parts of it. In my case I chose to write my own importing tooling essentially without learning at all about the built-in one: https://github.com/Julian/alubia. I had no intention to make that approachable for lots of users not named me (in fact none of my actual importers are present) but it's been very fun to watch my ledger get more and more accurate.

I've tried nix-darwin a time or two in the past. Every few years when homebrew makes a "hostile" change and I get upset I consider trying it again (now most recently with changes to gatekeeper). I think I'll get to doing so in the next year or so.

But I think just in fairness, the comparison here for flakes should be to Homebrew bundles. My packages are managed in a bundle: https://github.com/Julian/dotfiles/blob/main/Brewfile and then locked by a lockfile: https://github.com/Julian/dotfiles/blob/main/Brewfile.lock.j... and installing is just `brew bundle install`. All native Homebrew functionality. In practice I have never had an issue with non-reproducible builds across my machines (partly because the tendency on macOS is to run the latest versions of things and stay up to date).

(But again I do find nix-darwin interesting to try for other reasons.)

It's not trivial for a mathematician to understand Lean code, but it's something that's possible to learn to read and interpret in a day (without then necessarily being proficient in how to write it).

That's true though of Lean code written by a human mathematician.

AI systems are capable (and generally even predisposed to) producing long and roundabout proofs which are a slog to decipher. So yes the feeling is somewhat similar at times to an LLM giving you a large and sometimes even redundant-in-parts program.

What are you referring to? Inverting scroll wheel direction in macOS is trivial (and one of the first things I change), you just uncheck the "natural scrolling" checkbox.

I found this a few months ago and use it for something totally unrelated to containers -- namely on my Tailnet I have machines on my home LAN which I want to connect to even though they are asleep. I haven't been bothered to learn whether I can set up Wake on Lan requests triggered from my router to get them to wake up on demand even when I'm not on the actual LAN, so instead, when I want to connect to them, I use wait4x with a TCP connection to wait till the machine wakes itself up momentarily every few minutes, and then finally run what I want (which is usually either git pull or sshing into the machine).

> It's a shame that no browser cares to integrate that

I'm trying out using the Obsidian Web Clipper extension, which essentially does this (and using it for anything I'd previously have bookmarked).

Not directly relevant, but my PTAC in NYC started having issues this summer just as things got hot (of course).

The compressor would come on for a few seconds then shut off.

After 2 different HVAC companies quoted me $275 to come out (plus hourly and the repair once they find the issue) and then also told me it would be 10 days before they had availability I finally bit the bullet, bought a $30 multimeter, watched a few videos on how capacitor failure is super common and how to hopefully not kill myself, and after confirming with the multimeter and buying the $7 capacitor everything was right back to working with 2 minutes of work.

I did have a moment where I dreaded thinking I'd need to replace the unit and if so whether I'd want a split put in but for $53K I'd better get a third job... Quite glad not to have had to get too far down this road.

Always a take which sprouts vehement arguments, but this is something macOS gets extremely right, as Cmd-C, Cmd-V, etc. indeed work everywhere including the terminal, and Ctrl-C in the terminal is reliably SIGINT.

As can the (FOSS) LuLu: https://github.com/objective-see/LuLu

I do, it's given as a seminar twice a year.

Looks very nice!

I have an interactive tutorial I wrote and teach with which is here: https://github.com/JulianEducation/CommandLineBasics in case it's useful as well. I only have 90 minutes in my case so it's a constant battle to tweak what I can get to with my audience, so there's still lots of things I want to change.

But I think it's very important to have lots of resources here so I'm excited to look at yours.

To me GNU parallel is the top of the line here (as it is for most things parallel).

For the most common cases I have it aliased to just `p`: https://github.com/Julian/dotfiles/blob/main/.config/zsh/com...

Or https://github.com/Julian/dotfiles/blob/4d36e6b17e9804a887ba...

Mine is here [1], I basically had all of these set already (other than column ui which I don't like), but I suspect that's just because I've probably previously read nice posts by Scott and others talking about them.

Maybe trading alias tips is another useful thing to do though, hence sharing the link.

[1]: https://github.com/Julian/dotfiles/blob/main/.config/git/con...

> shouldn't proof checkers have a database of theorems and be able to fill in the intermediate steps or confirm that it's possible to fill in the missing steps?

This is essentially exactly what Mathlib[1] is, which is Lean's database of mathematics, and which large portions of the FLT project will likely continually contribute into.

[1]: https://github.com/leanprover-community/mathlib4/

(Other theorem provers have similar libraries, e.g. Coq's is called math-comp: https://math-comp.github.io/ )

Considering that part of the point is being able to use end-of-life devices, it's doing this in Termux that would be silly, don't run an OS which isn't getting security patches.

It's in at least 3 places on the homepage -- in 2 screenshots and in the first main code block you show. This was the first thing I noticed too, so I think it's good feedback from OP.

I haven't tried to use this, but if someone has, how does it compare to alot (https://github.com/pazz/alot) which is the best TUI mail client I've used?