HNHacker News
TopNewBestAskShowJobs

perrygeo

3,193 karma · joined April 28, 2014

submissionscomments
perrygeo··on US Department of Justice has officially reclassified cannabis as less dangerous
I am thankful to live in a town with a full ban on public smoking; weed and tobacco are legal to buy, just not to smoke or vape in the city center.

It's a clear and rational line - do unto yourself if you must, but keep the air clean and don't force others to share your nasty habit.

perrygeo··on I don't want your PRs anymore
As it should be. Issues are a dime a dozen, sometime coherent, rarely relevant. There is no barrier to submitting a garbage issue.

PRs, at the very least, provide a quality gate. You have to a) express the idea in runnable code and b) pass the tests.

perrygeo··on Salmon exposed to cocaine and its main byproduct roam more widely
Carter Vail was right. Cocaine was invented by salmon - I bet you didn't know that. https://www.youtube.com/watch?v=GX0tVk5T6zc
perrygeo··on PgQue: Zero-Bloat Postgres Queue
If you're looking for kafka-like semantics, you might want to keep messages around.

Your temporal partition idea is spot on. But instead of dropping old partitions, you can instead archive them.

perrygeo··on Official Clojure Documentary page with Video, Shownotes, and Links
You might be interested in https://jank-lang.org/ - still early days but full C/C++ interop is the plan.
perrygeo··on Ask HN: Why Hasn't Clojure Caught On?
As I tried to explain above, Clojure is made for a specific set of problems. Your problem seems to be interop with the SDK of the month and hiring. My problem is mutable state. It's logical that we'd choose different solutions.
perrygeo··on Ask HN: Why Hasn't Clojure Caught On?
Some thoughts:

1) Clojure is stable and there will never be big news or big changes to the core language. It's simple in the objective sense. There's fewer patterns to learn, hence less to talk about online. The code I wrote 12 years ago still works. The books I bought 12 years ago are still relevant. To an amateur github star gazer looking only at the metrics from the past month, this looks like stagnation. To me, this looks like good engineering.

2) The can-do pragmatic attitude meshes very well with entrepreneurs and small teams writing proprietary applications that need to get things done. These are NOT the people evangelizing and marketing open source tools. Clojure's successes are private, small, and quiet. In general, there is little to no focus on external validation.

3) Clojure is unabashedly a tool for experts. Don't get me wrong, the community is amazingly welcoming to newbies (as I discovered). But in order to align with Clojure's value proposition, you need to understand the problems it solves and feel them deep in your bones. If the words "mutable state" mean nothing to you, Clojure is going to feel wierd.

These conspire to make Clojure less visible online. Clojure's core audience, expert programmers who focus on outcomes and stable code, they do not read or write SEO blog spam.

The trending technologies, those that change so much they require articles like "How to do X in Y in April 2026" are built on shaky foundations. Trending means churning. That's hardly a value worth chasing.

perrygeo··on Modern Common Lisp with FSet
Hidden 13 paragraphs down the third page is the first actual description of the project:

> So FSet has a dual mission: first, to bring Common Lisp up to date, by giving it a much richer ensemble of functional collection data structures, to greatly expand the space of algorithms that can be written in an elegant functional style and still run efficiently; and second, as with Clojure, to support and encourage the use of functional collections for general programming.

Cool project but the docs could be greatly improved by putting the purpose of the project front and center. Don't make readers guess.

perrygeo··on Arguing with Agents
It's more than context-rot.

If you ask a vague ignorant question, you get back authoritative summaries. If you make specific request, each statement is taken literally. The quality of the answer depends on the quality of the question.

And I'm not using "quality" to mean good/bad. I mean literally qualitative, not quantifiable. Tone. Affect. Personality. Whatever you call it. Your input tokens shape the pattern of the output tokens. It's a model of human language, is that really so surprising?

perrygeo··on Astronomers Propose New Way to Find Aliens Without Knowing What It Looks Like
Assembly theory approaches the question from a different angle but largely reinforces this conclusion. Essentially, anything past a significant level of chemical complexity is a sign of life. DNA is astronomically unlikely to form spontaneously. It was designed by co-evolution. If you find a screwdriver on an alien planet, you can deduce the existence of screws, machines that require fasteners, and ultimately a living being that built those machines.

In the history of the universe, we only know of a single instance where something ejected from planetary atmosphere by its own force. Humans, when we launched a rocket into space. That's it. Debris enters planets atmosphere but we never see a rock LEAVING an alien planet on its own power. Except on earth.

A planetary system with evidence of panspermia would be a similar phenomenon to DNA, a screwdriver, or a rocket - clear evidence that life was at work to produce something otherwise so implausible.

This works so well as a framework for finding alien life since it makes no assumptions about what life is or what it consists of. Only that it produces complexity.

perrygeo··on Ask HN: We dont need a programming language now?
Current LLMs are powerful because they are trained on the patterns of existing data.

Without training on real token patterns, what outputs would you even expect to see? It's an LLM. Literally, a model of language. A language is necessary, by definition, even if you have to conjure it within your context window.

You're asking something closer to: "if LLMs can handle ANY language, why not assembly or something low level, close to the metal as possible?" That's a valid question.

My initial response is two-fold:

1) because someone still needs to verify and debug the source code - so human readability still matters.

2) because semantics are king. LLMs work because they can suss out relationships between tokens. At the lowest level, hardware doesn't really operate on tokens - moving 8 bytes from one address to another carries no information about what those 8 bytes represent to the larger program. Higher-level languages allowed humans to form richer semantics, just as they allow LLMs to express richer semantics with fewer tokens.

perrygeo··on FDA to weigh easing limits on unproven peptides favored by RFK Jr
The thing is, we have evidence that these "unproven" peptides are potentially very effective in certain cases. We just don't really know if they're safe universally. The FDA needs to make sure it's both safe and effective.

One reason to support this move: by regulating the peptide markets, the FDA could make it significantly less risky. Otherwise, people are going to hear the studies and buy BCP-157 from god-knows-where. There is a ton of junk on the market and it's a public safety concern. The FDA taking an active role in approving (or not) would be most welcome. Letting an semi-legal grey market thrive without any input from the government is a complete failure of the FDA.

perrygeo··on Do you even need a database?
Let's put it this way. I always end up needing the functionality and ACID guarantees of a database. I always wish I had a database. But some times I'm forced to use the project's legacy data stores (often flat-file data lakes) and watch every wheel get reinvented as we struggle to glue consistency, transactions, a bespoke query language, etc. onto an unwilling pile of unstructured data.
perrygeo··on Understanding Clojure's Persistent Vectors, pt. 1 (2013)
Most languages have some sort of peristent/immutable data library available. But that's not the same as having it built in! If you have to weave 3rd party types through the entire application and convert on behalf of libraries that aren't compatible. There is an enormous benefit to having these data structures as first class citizens.

And thank you for highlighting the conceptual link to Rust's borrow checker. Clojure and Rust are far and away my two favorite languages and this largely articulates why: they both have language-level constructs to completely solve the data ownership problem. Not just "here's some tools to optionally solve it, good luck" but truly, fully solve it. Full stop.

Almost every other language leaves ownership enforcement up to the developer. Clojure and Rust take entirely different approaches to virtually everything, but they converge on this one point and it's a crucial one. Types, syntax, compilation models, garbage collection, etc. are all secondary concerns to managed data ownership IMO.

perrygeo··on How to breathe in fewer microplastics in your home
What amazes me about the plastic health debacle is that any rational decision making process would have identified these compounds as entirely novel, with no evolutionary correlate. The precautionary principle should have applied; the burden should have been on corporations to prove these compounds safe. Instead, they were incorrectly assumed to be safe, deployed at scale, and now the burden is on us to deal with the impacts.

This was an experiment run without consent, and without accountability. Much like our experiment with atmospheric CO2, these companies are making profits by dumping the liabilities onto the public.

perrygeo··on AWS engineer reports PostgreSQL perf halved by Linux 7.0, fix may not be easy
Yet we're talking about postgres, specifically. The whole point is that benchmarks about postgres better know how to configure postgres or their conclusions be irrelevant at best. What does redis have to do with this discussion?
perrygeo··on 1,200 years of peak cherry blossom dates in Kyoto
I love this chart. Climate education gold. People don't consciously register a .2C rise in temperature or a 2mm rise in sea level, but they notice when things bloom 2 weeks early!
perrygeo··on Maine is about to become the first state to ban major new data centers
I understand the resistance to data centers. They have local environmental impacts, raise energy prices, and generally funnel wealth away from the area. But that's not inherit - it's the result of policies and subsidies.

Instead of banning, why not put the REAL price tag on it? Require local employment. Stop subsidizing infrastructure. Start taxing them. If a data center came in and provided jobs, paid their own way without handouts, and lowered electricity costs, it might be a different story.

It's not data centers per se. It's the crony capitalism that subsidizes them and lets them extract wealth without oversight. Regulate it properly and the negative characteristics go away. And yes, putting the actual price tag on it might scare away most investors - an effective ban. But that's the market working efficiently on price signals.

perrygeo··on Clean code in the age of coding agents
This article is more about how to get LLMs to adhere to existing definitions. I was hoping this would explore some re-definitions of "clean code".

DRY is a principle that comes up frequently. But is repetition really that bad when LLMs can trivially edit all instances of the pattern and keep them in sync? LLMs, by contrast, cannot understand a leaky abstraction - the typical result when you hastily apply DRY. So "clean code" in a era of LLMs might be mean more explicit and repetitive, less abstract.

perrygeo··on Old laptops in a colo as low cost servers
I don't see it. Hobby projects can use a VPN tunnel to make a data center from local equipment. Real projects that choose colocation have uptime requirements that simply can't be met by random consumer hardware. The venn diagrams don't intersect.

There's no middle ground where you try to run a real business on old laptops. That's insane. You either keep things small/hobby and stay simple, or graduate to production-grade equipment once you have real requirements.

The middle ground, taking on production colocation problems plus the unreliability of random hardware, sounds like the worst of both worlds. There are both simpler and more robust options.

perrygeo··on Old laptops in a colo as low cost servers
Old laptops as low cost servers? Absolutely, build a homelab in your own basement, rent a cheap VPS, set up wireguard and viola - instant data center for tens of dollars per month. It's not production grade but you'll learn a ton.

But colocation?

Strip away the learning component and add production uptime requirements - why would you even consider using crusty old laptops for this? If you have production grade needs, look to a standard cloud provider or, at the very least, a colo facility where you can put production-grade equipment.

perrygeo··on What if the browser built the UI for you?
Circa 2008 I built an internal web framework (based on Django) with the goal of defining everything as server-side models; the client/presentation layer was derived entirely from the model description. Zero front end programming.

A couple of notes:

- It works, technically. This ain't new. CRUD UIs can and have been totally automated. It was well within any competent programmer's grasp 20+ years ago.

- It doesn't work from a business perspective. Everything looks stock, generic. No branding. No UI clues to guide behavior. Business objects felt siloed, their relationships unclear. You might as well manage your data in a spreadsheet.

Ultimately, the UI needs turn out not to be a straight reflection of the data model. Every single project required additional judgements, joins, aesthetics, branding, custom workflows, etc. Thus every project eventually ditched the auto-UI and built a SPA-style javascript app against the REST API.

In short, generating a UI from a data model is an old idea. The reason we don't use it today is that it sucks.

perrygeo··on AWS engineer reports PostgreSQL perf halved by Linux 7.0, fix may not be easy
Honest question: what's the value of running the benchmark and reporting a performance regression if the author is not familiar with basic operation of the software? I'd argue that not understanding those settings disqualifies you from making statements about it.
perrygeo··on Clojure: The Documentary (April 16th) [video]
Clojure has the best origin story of any programming language. Makes me think: If everyone took a sabbatical and worked on their dream, the world might look quite different. Thank you Rich for making your dream open source!
perrygeo··on Thoughts on slowing the fuck down
Wonderful analogy.

I still use spreadsheets regularly. Relieved of the pressure of making something "good", I can get basic things done quickly.

Is it sustainable or maintainable? Nope. Doesn't need to be. Not qualities I'm remotely concerned about when I'm writing a spreadsheet.

Vibe coding is similar in that you can solve a specific problem without much concern for generalization or future reuse.

perrygeo··on Ask HN: Did you ever consciously create technical debt?
The original talk by Ward Cunnigham made this clear; a conscious choice is required for it to be considered debt. You are explicitly releasing something you KNOW is suboptimal, specifically because you want the benefits today and have anticipated the costs to pay it down later, as well as the interest payments should you chose not to. You're using financial leverage the same way as you would buying a house or car.

By that definition, we do this all the time. I'd wager every feature release has some degree of "oh we can address that later if/when this takes off".

If you accidentally introduce bugs or regressions that gums up the works, that's not "debt", that's a mistake. If you choose the wrong thing and realize too late, that's not "debt", it's just bad decision making. If you choose wrong and are dissallowed from ever going back and cleaning up as agreed, that's not "debt", it's bad management.

We've got to stop using "tech debt" to mean "everything we don't like about software".

perrygeo··on Thoughts on slowing the fuck down
> What are you building?

This x1000. The last 10 years in the software industry in particular seems full of meta-work. New frameworks, new tools, new virtualization layers, new distributed systems, new dev tooling, new org charts. Ultimately so we can build... what exactly? Are these necessary to build what we actually need? Or are they necessary to prop up an unsustainable industry by inventing new jobs?

Hard to shake the feeling that this looks like one big pyramid scheme. I strongly suspect that vast majority of the "innovation" in recent years has gone straight to supporting the funding model and institution of the software profession, rather than actual software engineering.

> I'm not even sure building software is an engineering discipline at this point. Maybe it never was.

It was, and is. But not universally.

If you formulate questions scientifically and use the answers to make decisions, that's engineering. I've seen it happen. It can happen with LLMs, under the proper guidance.

If you formulate questions based on vibes, ignore the answers, and do what the CEO says anyway, that's not engineering. Sadly, I've seen this happen far too often. And with this mindset comes the Claudiot mindset - information is ultimately useless so fake autogenerated content is just as valuable as real work.

perrygeo··on More common mistakes to avoid when creating system architecture diagrams
Most common mistake: Creating diagrams in a graphics program without semantics. Why not write your "diagram" in code or at least some structured format?

- LLMs can read and write it, it serves as valuable context

- Always up to date, rendered on demand

- source controllable

- editable (thus will actually get edited)

Most architectural diagrams I see are just images with no source or semantics. Strictly less useful and more prone to rot.

perrygeo··on Why I love NixOS
Easier and largely compatible with the rest of the world. Solving problems with "If we all switched to NixOS..." is a non-starter in most organizations.

My rule of thumb: keep a strict separation between my projects (which change constantly) and my operating system (which I set up once and periodically update). Any hard nix dependency inside the project is a failure of abstraction IMO. Collaborating with people on other operating systems isn't optional!

In practice this means using language-specific package management (uv, cargo, etc) and ignoring the nix way.

perrygeo··on The bespoke software revolution? I'm not buying it
The default has been pay $x/month for every service. I've seen startups that require a dozen service accounts just to run the software, and dozens more to get onboarded org wide. One service for feature flags. One service for logs. One service for traces. One service for error handling. Another service for ticket tracking, which is completely separate from your planning, design, and CI services. Jesus. What do people hope to accomplish here besides just defering blame?

Replacing SAAS isn't about building a replacement services 1:1. It's about figuring out what you actually needed in the first place! Often we only use a tiny fraction of what the full-blown SAAS offers. IOW it's about eliminating the service entirely and building something that fits your actual needs, rather than following what some VC thinks your needs are.

AI or not, the "build vs buy" pendulum is now swinging hard to build. And IMO that's a real opportunity to consolidate, trim some fat, and actually apply engineering practices rather than just blindly signing up for every SAAS that crosses your path.

← PreviousPage 6 of 34Next →