Hacker Newsnew | past | comments | ask | show | jobs | submit | ruszki's commentslogin

> I'm not a fan of APIs that surprise.

It really depends on what you got used to. C++ was the first language, which, I would say, learned. I’m still surprised to day that traversing a set is not in order by default.

In this specific case, emplace should be your default option, and you should really know why that’s the case, and why you have this many options.


Were the tests generated by an AI then? How do you know whether they are really comprehensive?

Yes, effectively my entire project was generated by AI.

It's a very weird and uncomfortable way of working - I've said in the past that I don't like a single line of unreviewed AI-generated code in anything beyond a prototype, and now here I am with 13,000+ lines of mostly unreviewed Python written by Claude Opus 4.5.

I'm leaving the alpha label on it until I'm a whole lot more comfortable with the codebase!

I do however know that the tests are pretty comprehensive because I had the model use TDD from the very start - write a test, watch it fail, then implement code to make it pass.

I was able to keep an eye on what it was doing on my phone while it worked and the TDD process seemed to be staying honest.

Here's one example from the full transcript, showing how it implemented closures: https://static.simonwillison.net/static/2025/claude-code-mic...


I'm now having Claude Code build the tests for my voicenotes organization application. For the most basic implementation - just a single text field - I wrote in English which tests I know I need, there were about two dozen. Approaching size limits, unicode, normalization, nonprinting characters, Hebrew vowel points, empty strings vs NULL strings, Exceeded byte length without exceeded character length, etc etc. I then threw Claude Code at it.

Claude Code found more edge cases to write tests for than I ever would have thought of. And I've been doing this for 20 years.


The leetcode style tasks have nothing related to coding for the past 2 decades. That part is solved for a long time now. They ask for knowledge, which is a search away from everybody. I don’t know anybody, who knows those, and not only because of interviews. Also interviewers ask these, yet average code isn’t optimized at all. A simple question, like what’s your opinion of <anything> will tell you more than any leetcode question.

Except of course these things:

- Reliable internet sharing, especially when connection is spotty, and when your connection switches between operators or countries

- Making alarms randomly silent. I missed a flight once because of this. There is no excuse for this.

- Randomly not working AirPlay

- GPS is terrible compared to any of my previous Androids. Even my first Android in 2010 was better than this.

- Finding an operator can take a loooong time after crossing border

- Random restrictions in App Store, like no torrent clients

- Generally terrible keyboard for my native language (Hungarian). Prediction and basic accent fixing doesn’t work at all. The exception is when I don’t need to change a word with diacritics… when the keyboard’s dictionary clearly contains them

- Apple Maps is still a joke. Many times it doesn’t load the underlying map layer at all. I switch to Google Maps search for what I want, finding it, reading some info, looking some images, then switching back to Apple Maps, and it still doesn’t load. Also, navigation and speed limit information are unreliable to say the least.

- Heavily underdocumented MacOS virtualization API, and half the features can’t be used in a real environment, but these restrictions are completely undocumented

- Wanting to have a running DNS server is a challenge on MacOS

- Unusable GPU when no monitor is connected

- You basically need to turn off all security features in MacOS to allow some basic automation, like with FaceTime

- Generally terrible compatibility with anything non Apple. Do you want to show your photos on your friend’s random TV without hassle? Good luck.

- Many built in features (eg SSH, VNC) are heavily restricted, and good luck if you want to replace them cleanly. Most information on internet is “just use the built in solution”. Also they are many times completely insecure.

There are hacks for these, but “they just work” is not true at all. On the level of how “they just work”, top level Android and Windows devices are also on that level for more than a decade in case of Android, and at least 20 years for Windows (if not more). Maybe Apple TV is my only device which just works without hitting some quirks. Especially compared to my other TV and TV adjacent devices. But even here its FaceTime solution, let’s say “interesting”.


They won’t, but Apple previously lied similarly against PWAs.

You can reimburse your costs of unplanned overnight stay, even when it happened because of weather. So those AirBnBs were also free. Even the taxi to and from there. Ryanair was unlawful if they hadn’t given this information.

Btw, there are power banks and headphones which can easily handle 2 days.


Surely you can recharge them on the ground too. Do Ryanair not provide in seat usb sockets? Wouldn’t surprise me if they saw an opportunity to charge £5 for their use (log into wifi and activate them etc).

It’s not that simple. I’ve seen several routes where ultra low cost carriers simply lost against normal airlines. Especially when there are multiple airports in a given city, they often loose at usable airports. For example in Brussels.

Especially that it’s completely normal to talk with each other on planes.

Could? I flew a month ago on a flight with Starlink. I downloaded 10s of gigabytes of data without hiccups. Calling was not an issue. And it was completely free.

Quite curious, which route was this on?

Not one time have I had a consistent internet connection whilst flying transatlantic on Delta, KLM or BA airplanes, to the point that I regretted having paid for it every single time.


I haven’t been to India for 7 years but I distinctly remember a very productive flight from Delhi to Heathrow on wifi while I was sshed into a machine and working on something for hours with no issue - far better than the signal I get on the train from London to Manchester.

Qatar/Virgin Australia, Sydney->Doha. I’ve never had really good connection either before this, and I tried many-many times. That was the exception when it worked as intended.

When we collected, correlated, and measured all controlling messages in a whole 4G network. Millisecond precision meant guaranteed out of order message flows.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: