Greg Lorriman @GregL2

I can’t believe it’s 2025 and we still train junior devs to write hundreds of unit tests before writing a single integration test Nothing burns more time and makes refactoring harder than mountains of pointless unit tests

I've written a shorter tl;dr article on the misunderstanding of exceptions, also by library writers, that has people catching them in calling code, replicating the clunkiness of error-codes and defeating the pupose, on my linkedin. Below with a poll. Tell me what you think.

I hate the unnecessary bother and clutter of error-codes. I've written a shorter tl;dr article on the misunderstanding of exceptions, also by library writers, that has people catching them in calling code, on my linkedin. Below with a poll. Tell me what you think.

I find the lack of exceptions in Swift lamentable; it's making me think to go to Kotlin; I hate the unnecessary bother and clutter of error-codes. I've written a shorter tl;dr article on the misunderstanding of exceptions on my linkedin. Below with a poll. Tell me what you think.

Where am I going wrong? There's a drift against the use of exceptions and I think it's a mistake. Swift, for instance, doesn't have real exceptions at all. Does my article make sense? See the link 🔗 below.

Makes a lot of sense if you know how they work. Still, if there's a website or book or reddit post that lists what might be missing as "things people forget" or some such, then the LLM might be able to spot it.

BURKOV @burkov

3 months ago

Makes a lot of sense if you know how they work. Still, if there's a website or book or reddit post that lists what might be missing as "things people forget" or some such, then the LLM might be able to spot it.

4 12 75 6K 58

Download Image

As weak as AI is in reality, if it’s 1% better than 80% of employees, it can replace them. It’s similar to self-driving cars: if self-driving car kills less people than human drivers, then it’s a go. x.com/erikbryn/statu…

Erik Brynjolfsson @erikbryn

2 years ago

As weak as AI is in reality, if it’s 1% better than 80% of employees, it can replace them. It’s similar to self-driving cars: if self-driving car kills less people than human drivers, then it’s a go. x.com/erikbryn/statu…

58 191 984 1.2M 149

Download Image

If you think about how LLMs work, this was the only conclusion. But there has been a lot of evidence for some time also. It was logically impossible that LLMs would ever reason.

Ruben Hassid @RubenHssd

4 months ago

If you think about how LLMs work, this was the only conclusion. But there has been a lot of evidence for some time also. It was logically impossible that LLMs would ever reason.

3K 9K 65K 14.1M 42K

Download Image

The hardest part in LM-assisted coding is to stop. The process is crazily addictive, especially when you build something you always dreamt of.

CS graduates are often no better than bootcampers. Maybe except if the uni was a Russell group university. I helped a non Russell Group undergrad. She was utterly useless and got a 2:1!!???! I remember teaching hashtables at one of my jobs. Basic stuff. youtube.com/shorts/ehv-Ls9…

Reminder: this is all fake. Any time someone makes a claim like this, either they are lying, or every single programmer they know is completely awful and can’t program at all. x.com/alexalbert__/s…

Alex Albert @alexalbert__

4 months ago

Reminder: this is all fake. Any time someone makes a claim like this, either they are lying, or every single programmer they know is completely awful and can’t program at all. x.com/alexalbert__/s…

192 167 3K 975K 1K

OMG, o3 is poor at guiding you through a project setup. Asked it for a detailed step-by-step to bootstrap a Svelte web application on a Mac: errors on every command I run. I fought with it for half an hour and switched to Gemini. No issues at all. All commands resulted in…