Been doing some architecture work lately, and decided to start rendering these using Factorio sprites and factory style layouts. For me, it's been extremely useful for spotting certain patterns and problems.
Released as open source, feel free to take it for a spin. PRs are welcome.
As someone who has been using alot of spacetime for side projects (like https://heat.echohack.app), I am continually impressed with the speed at which it operates.
I think there's alot of interesting things happening in the database space. Vitess/Neki, vector stores, spacetime are all really good things to be happening. I think it's a shame that database developers seem to have a drama filled timeline out there. It's... all very exciting, together.
Anyway, Some things that need improvement (some of which are addressed by this blog post):
1. Backups (fast recovery) and Disaster Recovery (slow, durable recovery)
This is a big one, but sometimes speed is not the only objective you have to meet. You need to have certainty that, if everything goes down that you (eventually) can bring things back online. I don't really have a way of doing that today.
2. Read replicas sure would be nice for analytic workloads
3. Durable writes to s3 would be nice for intermittent bursting workloads (like ci systems)
I think the Spacetime folks have their work cut out for them, not necessarily because of the technology (that's hard too) but because the AI models have seemingly decided that Neon and Postgres are all that exists.
I think spacetime has a bright future ahead of it, and I am wishing the team all the best as they work hard to imprint something new on the universe.
The weirdest payment method has got to be the Coke ON vending machines that do not support SUICA but instead require you to QR code-IR blaster you to a website where you bluetooth pair with the vending machine then buy it on your mobile browser.
Like....somehow they turned a sub 1ms operation into a 4 minute 32 second one. It boggles the mind
Most Coke On machines do support Suica/Felica but you have to press IC on the blinking pad. The problem is that the pad doesn't seem like a touchscreen and it can be difficult to precisely tap. However Suntory does have app-only machines now.
It's not really mentioned in the article, but Suntory and transit operators out in Hiroshima etc. are turning away from Felica simply because the licensing fee is too high for them. That's also why Samsung and Google don't enable it on non-Japanese phones. Sony and JR missed the part where they have to generous first to eliminate all competition before jacking up fees, and now they're getting eaten by QR payment and Visa.
Coke ON Pay uses an app that automatically pairs with the machine when you open it [1]. It’s pretty quick and takes less than 30 seconds once you set it up.
I think you might be referring to QR de Kessai [2], which was introduced in 2023 and is targeted more at inbound international tourists who don’t want to download the app and may not have an E-Money card. Basically they aim to support to support the widest possible audience - at the cost of a slow/clumsy experience. These machines should also accept payment via the faster Coke ON app method and often have E-Money tap-to-pay options too.
From that perspective it kinda sorta makes sense. But yeah I wouldn’t bother with it unless I was dying of thirst and I had no cash or other payment options available.
30 seconds is the complete user flow to open the app, select a drink, approve payment with face id, and dispense the drink. Maybe a few seconds longer in the end than paying with Suica, but pretty negligible.
And a lot of coke machines in train stations, service areas, etc also accept Suica if you prefer (but then you don't get the stamps).
Typically the machines I see which only accept the Coke ON app are in more rural areas that traditionally would have only accepted cash. And I much prefer the app to cash.
In the US, if you do not have a passport and want to apply for a new one, you have to go to certain places where an authorized person will check your paperwork, see that you match the paperwork and send that directly to the passport office. Generally, county clerks and large post offices are the places that most people go. In some states, libraries are also authorized to do so. In 2026, Trump signed an executive order banning libraries from doing so. Libraries were doing the service for free. County clerks in my state charge around $50 for the service. The libraries doing this service were almost always in "blue" districts.
The SAVE ACT [0], which passed the House of Representatives last year (and is stalled in the Senate) only allows 4 types of identity documents to be allowed to register to vote in federal elections:
1. A US Passport,
2. A US Military ID with proof of US citizenship.
3. A US REAL ID with proof of US citizenship [1].
4. A US REAL ID without proof of US citizenship, plus a US birth certificate and the names must match [2].
1 - This is called Enhanced Driving License. It exists in only 5 states (Michigan, Minnesota, New York, Vermont, and Washington) and is an additional $35-40 above the cost of a regular driving license.
2 - In the US, it is common for a wife to take her husband's surname. This means that 74,000,000 US women will lose the right to vote in federal elections if they don't spend $150 (for a new passport) or hundreds-to-thousands of dollars changing their name legally.
You don't need to make libraries illegal to get rid of them, you just need to undermine their value and they'll be defunded and disappear. No passing laws necessary, just pubic ambivalence.
The oxford comma debate is so annoying because it clearly has nothing but advantages. Removing commas from a delimited list does nothing to resolve ambiguity, whether lexicographically or syntactically.
It's so useful as a delimiter and anti-ambiguity machine, that you don't even need spaces for it to work! See CSV or Japanese.
I have found AI great in alot of scenarios but If I have a specific workflow, then the answer is specific and the ai will get it wrong 100% of the time. You have a great point here.
A trivial example is your happy path git workflow. I want:
- pull main
- make new branch in user/feature format
- Commit, always sign with my ssh key
- push
- open pr
but it always will
- not sign commits
- not pull main
- not know to rebase if changes are in flight
- make a million unnecessary commits
- not squash when making a million unnecessary commits
- have no guardrails when pushing to main (oops!)
- add too many comments
- commit message too long
- spam the pr comment with hallucinated test plans
- incorrectly attribute itself as coauthor in some gorilla marketing effort (fixable with config, but whyyyyyy -- also this isn't just annoying, it breaks compliance in alot of places and fundamentally misunderstands the whole point of authorship, which is copyright --- and AIs can't own copyright )
- not make DCO compliant commits
...
Commit spam is particularly bad for bisect bug hunting and ref performance issues at scale. Sure I can enforce Squash and Merge on my repo but why am I relying on that if the AI is so smart?
All of these things are fixed with aliases / magit / cli usage, using the thing the way we have always done it.
Because it's not? I use these things very extensively to great effect, and the idea that you'd think of it as "smart" is alien to me, and seems like it would hurt your ability to get much out of them.
Like, they're superhuman at breadth and speed and some other properties, but they don't make good decisions.
This was a fun thing to realize in my early days of programming in Delphi. I guess the author will soon realize why old systems name things ticket "00001" and so on.
Been doing some architecture work lately, and decided to start rendering these using Factorio sprites and factory style layouts. For me, it's been extremely useful for spotting certain patterns and problems.
Released as open source, feel free to take it for a spin. PRs are welcome.
- Echo
reply