Alexander Torstling

@Shiroi990032 @SmokeyStack_ Nice idea! Would the grizzlys be passive or hostile? And how would you interact with them? How would you obtain marshmallows?

@SmokeyStack_ I have two requests for this update the addition of grizzly bears and marshmallows ıt would make for a fantastic camping experience especially since the sitting mechanic and straw bed are already in the game Also instead of breaking the straw bed should transform into a straw bag

Everything makes more sense now. Why wool stairs of all things were the first blocks to be data driven. Mojang is slowly but surely getting both internal devs and creators closer to the same block authoring process so the two are using the same tech.

@SmokeyStack_ Yes, we've been working hard towards aligning vanilla and creator on block authoring. We hope that this will give more power to creators and a better workflow for vanilla. Win-win if it goes well. Archetypes are essentially legacy vanilla stuff yet to be properly componentized

@allenholub I agree in principle but think this requires some nuance. Depending on the software, you cannot always get to zero known defects. For instance, in games, systems like physics and graphics aren't perfect. What's more realistic is to triage all bugs and either reject or fix them.

The whole idea of "fixing technical debt" as an activity separate from programming seems wrongheaded to me, at least if you define tech debt as bugs. To me, the goal should be no technical debt to begin with. There should be nothing to fix. The one exception is debt that's discovered through learning. You do the best job you can, but later on, you learn something that tells you your original work could have been better. Fix that when it comes to your attention. Refactor the code to look as if you knew then what you know now. But known bugs? No. There shouldn't be any. A no-known-bugs-on-release policy leads to faster development time, easier-to-modify code, and basically costs you nothing, and there are many shops that have that policy in place, including some parts of Microsoft. Clean up as you go. My own personal experience is that the time required to fix a bug goes up logarithmically with the time that's elapsed since the bug was created. Newly created bugs are easy to fix. It's vastly more time consuming and expensive to fix them "later." Put into perspective, I'd guess that the time and effort it takes to fix one old bug is roughly the time and effort required to fix at least 100 new ones. Minutes compared to days. Finally, before the people who don't read closely and jump to conclusions spout off, note that I said "known bugs." There are always surprises. Fix them as they make themselves known.

@SmokeyStack_ I value an efficient, elegant and powerful solution over any superficial code metric like bad formatting, code organization or language feature usage. It's like a book; grammar is important, sure, but it's the story that matters.

@SmokeyStack_ Clean code is certainly one thing. But I'm more thinking of the bigger picture; is the solution elegant? Is there some code you could reuse for this? Should the code exist somewhere else? Did we maybe even already solve this? What's the algorithmic complexity? Is it fast? Etc

I hear so much talk about code quality nowadays. But how about solution quality? No linting errors, no crashes, correct formatting, right license header, code comments, and documentation. But a terrible solution. I'd rather have a diamond in the rough than a pig in a dress.

@allenholub Primarily 6. Lack of developer agency due to cross-team dependencies, yet an expectation on single teams to deliver independently, without cross-team coordination. I.e. tug of war. And 4, manifesting in gating and focus on building things right and not building the right things

Another ridiculous poll. What actually slows releases? 1) Building things nobody wants. 2) Unnecessary complexity. 3) Lack of effective communication. 4) A culture of centralized control. 5) Bad process. 6) Dependencies. 7) Not trusting the people who do the work. 8) Large batches. I could go on. None of the things in that poll are significant factors.

@vincentparizet @molecularmusing AI for code review is like getting a review from your annoying younger sibling - it will tell you obvious things as if you didn't know about them

@molecularmusing Have you tried AI for code review?

Most people know my stance on it, but it has to be said at this point: As long as Live++ is 100% under my control, there won't be any AI feature of any kind in it, and exactly 0% of its code will be written by any AI. Same for future releases of other tools.

@avrldotdev The bartender was an intern

A Java programmer enters a bar. He orders a "beer." Another Java programmer orders "beer." The bartender gives them the same glass to drink from.

@Funny_Anar It's a cheat.

On blusky Jeb said that using F3 to get information about your world is considered cheating because that not something a player can realistically do. What do you think about this?

DHH thinks grinding is dumb. What he told us on My First Million: - "I often hear entrepreneurs talk with such pride in their voice. Yeah, I grind. This is why I'm working a hundred hours a week." - "I don't want to do that. My life is far too interesting to waste it grinding." - "Grinding is the stupid shit you do in World of Warcraft when you're a peon."

@tsoding This will work, you just need a proportionally big enough icecream

Look, we wear coats during the winter not because the coats generate heat, but because they are thermal insulants, right? What if I wear a coat during summer while eating ice cream? The ice cream cools down my body, the coat isolates it from external heat. I think this may work

I’m happy to announce the release of a new open source 3D physics engine called Box3D. I’ve been working on this project for a few years now, but it represents over 20 years of experience writing physics engines for games. Read more here: box2d.org/posts/2026/06/…

Box3D has been announced?!?!?! box2d.org/posts/2026/06/…

Lock-free programming isn't hard because of atomics. It's hard because of memory reclamation. C++26 standardizes Hazard Pointers, allowing you to safely defer deleting a node until all concurrent readers are done with it.

I have this theory that if one person is seeing a bug, in most likelyhood almost everyone is seeing it. And in 99.9% of cases, people don't bother to report things they encounter. So I always take any bug report extremely seriously.

@codingspirit_de @StokeWillie Yeah it's a bit of a rough part of c++ i find; there's iterators and you can use template functions, but neither are really great.

@atorstling @StokeWillie But only as view.. it's not modifiable, with remove or append

In C++26, a standard "in-place" vector can provide contiguous storage for non-movable, non-copyable, non-default-constructible types.

meshoptimizer 1.2 is out! Featuring support for tangent generation, a significantly faster vertex decoder for Intel/AMD CPUs, several new smaller algorithms (cluster position compression, triangle filtering, clusterlod BVH) and more. GitHub stars and retweets appreciated!

@OlexGameDev That makes sense. I would still clean it up for the stats sake, though, since it seems to be quite prominent in the top of the list; application->gameworld->audiosystem->x3daudio

@atorstling It's max accumulated time of 2 seconds but that is across multiple threads where the whole build is 80 seconds of work. That is something but not enough for my main CPU. Maybe if I was building on a laptop more often I might.

Why I don't use STL: shorter build times. As my C++ code base crawls to 200k, it's great to know that I keep the build times shorter with one simple trick: don't use STL headers. Currently sitting at 181k lines. A full rebuild is 11 seconds on my beast of a CPU: 9950X3D.