Artillain ( Mr. Zerolith )

Zerolith happenings 05/12/26: ZL is ~80% into production release readiness and still accumulating features! Debugger got 10x faster in scenarios where it's slow, allowing realtime compressed logging of debug output in production at low cost. And it can still get faster-er later. Soon, the debugger can feed an interesting form of fully automated testing ( no test writing needed ), y mas Unfortunately zyeet and zcache ( blast processing features ) are still being held back so that we can dogfood and reliabilitymax those libraries. We also found some footguns that need removal. Sometimes you don't see them until a shoe gets grazed. But.. Preview Release 3 is imminent. We are now in the merging/testing phase. We will dogfood and continue to evolve PR3 on 6 apps for 4-6 months, and see what needs to break before we call it a production ready release. Because we want the design to be heckin' stable for people to build very serious, very fast things on. Also we have a test application written in L*ravel and ZL so you can see the difference. Coming soon! Our new contributor, Labba, wrote 40% of the code and me and him are turning out to be quite the tag team. We are pumped to release this!

@_jeroka Gotta love that over the top 90's Japanese game art!

@artillain Love the graphics.

Zerolith happenings 05/12/26: ZL is ~80% into production release readiness and still accumulating features! Debugger got 10x faster in scenarios where it's slow, allowing realtime compressed logging of debug output in production at low cost. And it can still get faster-er later. Soon, the debugger can feed an interesting form of fully automated testing ( no test writing needed ), y mas Unfortunately zyeet and zcache ( blast processing features ) are still being held back so that we can dogfood and reliabilitymax those libraries. We also found some footguns that need removal. Sometimes you don't see them until a shoe gets grazed. But.. Preview Release 3 is imminent. We are now in the merging/testing phase. We will dogfood and continue to evolve PR3 on 6 apps for 4-6 months, and see what needs to break before we call it a production ready release. Because we want the design to be heckin' stable for people to build very serious, very fast things on. Also we have a test application written in L*ravel and ZL so you can see the difference. Coming soon! Our new contributor, Labba, wrote 40% of the code and me and him are turning out to be quite the tag team. We are pumped to release this!

@peduarte How dare they put some visual character into anything

when i first saw spotify's new app icon, my immediate gut reaction was: wow that's pretty cool then some dude with a few thousand followers and said it sucked and in a blink of an eye everyone started hating on it i still think it's pretty cool

@ilyamiskov Thanks for the context. Yeah this industry needs more Luigi Colanis. When someone tells me they're a designer, i don't take them seriously by default due to the lack of balls on the part of the profession. Risk is necessary. It's the only way the art is advanced.

@artillain I'd say, you need to be aware of the rules and why they're there in the first place. But don't be afraid to break them. By rules I've mostly meant "safe choices", some UX patterns that have been tested and proved to work by other companies. Stuff like that.

Obsessiveness with rules in design is just an escape for those lacking taste.

So i fired Hetzner. - 4 emails later, i still cannot get them to accept any credit card i own; paying monthly with paypal is too annoying - The administrative interface has critically bad UI/UX - No way to turn off incessant maintenance emails They really need someone to holistically rethink the non-technical part of the customer experience. I'm willing to pay more for a lower friction experience. The hosting product is otherwise very good for the price you pay. But everything else is grating.

@tekbog I just had my first experiences with a SaaS company over there, they have a special brand of ineptitude i have not encountered yet other than dealing with Hetzner. Does really feel like they suffer government worker syndrome

trying to work with european companies it's like dealing with bureaucrats man

@Sparkie_light @etscrivner GUI is one of the few places that object oriented makes a decent amount of sense Yet people will still slather a backend in it

@etscrivner It really seemed to take off with low level GUI where it still makes an incredible amount of sense. Now that C++ is not used for UI as often people are starting to question why they’re still using it.

OOP was (at least partly) an attempt to create a mechanical system for modeling any area of knowledge with only partial understanding. As a result, its default result is usually extremely defensive and verbose code (see screenshot). Techniques like abstract base classes and inheritance hierarchies are precisely for guarding against future changes caused by ignorance or lack of planning. The problem is this level of generality/flexibility is almost always unnecessary and has non-trivial compile time, run time, and complexity costs.

@etscrivner I can typically cut the code in 1/3 by refactoring average OOP down to procedural or light OOP and it will perform 5-10% faster, consume much less ram, and be easier to work with. I cannot identify a single positive trait of medium to heavy OOP

@round I do if it's good

nobody cares about your design process

@HSGARAGE1 I like EVs because: - instant acceleration - low center of gravity - if designed correct, very little can break or needs maintenance - Extremely efficient - I could produce the "fuel" myself if i invested in solar panels and battery array. Don't care for the noise of a gas car.

私はガソリン臭くて、燃費が悪くて、音がいっぱい出る野性味あふれる車が大好きだから旧車に乗ってるわけなんですが… EV車乗ってる人は何で買ったの？

@tonyennis @htmx_org I select for grug-compatible mentality when hiring at my shop and some of us ( even me ) still occasionally over-engineer things and have to dial it back. I think there is a factor of 'if all you have is a hammer, everything looks like a nail'

I'm building the PHP backend equivalent of HTMX, and here's my theories on why that happens: 1. In corporate software environments, which are the most common today, there are strong pressures that cause programmers to chose job security enabling technologies over simple ones that save work 2. Making hard things easy is one of the hardest things a programmer/designer could do, and is also a dangerously anti-job-security move in large corporate environments 3. What Neal Ford said

An idea that's been occupying me a lot lately: "*Low Status* tech/software practices". There are a lot of places where older or "dumber" approaches are actually better or faster, but where 99% of engineers won't acknowledge or even engage with them. cc @htmx_org

@tonyennis @htmx_org Notice that 'low status' exactly correlates with 'cost and time efficiently achieving the thing'

Examples of low status practices - Using a simpler stack so you can use "dumber" but faster models - Loading raw, no-build step CSS files straight from a *folder* (gasp) - Being ok with large PRs, using Claude to assist with code reviewing large PRs. - Coding on a VM (becoming less low status recently)

@plainionist I think it can help with solving truly novel problems, especially during the research/prototyping stages. I would not trust it to fully implement the novel technology, you would miss insights you'd get during the process that improve the result. You should be very involved.

“AI cannot solve truly novel problems.” True or false? 🤔

@Andercot Multiple generations of humans haven't wanted to strap the computer on their head. There isn't a single success story in this product category. What i wonder is what gave big corpos the confidence to try it again?

The absolute non-takeoff of VR and AR is probably one of the big upsets in consumer electronics history Pretty much everyone thought this would be huge and it sort of just isn't

@dipakcgajjar Local AI support? can it happen via the openAI protocol? ( user can set endpoint, username, password )

🚀 What’s new in WordPress 7.0? • Native AI integration (OpenAI, Anthropic, Google) via a unified Connectors screen • New Breadcrumbs, Icon & responsive Grid blocks • Customisable navigation overlays • Redesigned admin UI with smooth View Transitions • Iframed post editor & PHP-only blocks for developers • Block visibility by viewport + custom CSS per block WordPress just entered a bold new era. 🤩

Still think about this

@big_duca We started out with ~32b models on the 5090 but they just weren't smart enough even today for hard senior level coding work. GPT OSS 120b, i'd put right in the middle of a 30b and my 197B favorite, the only problem is that it's not great with agentic and can be hallucinatory

@big_duca Yes. Step 3.5 Flash 197B is a beast and has big model energy. Still ask deepseek hard research questions from time to time though, but for coding, step kicks ass and is very detail oriented. It was expensive to get the hardware to run it, but we did it.. RTX PRO 6000 + a 5090

Is anyone successfully using open source AI models to code? And if so, what and where can I learn? Feels like in the next few months the model companies are gonna start shafting us all hard.