@BenjDicken @nathanjmorton_ I would honestly make that a bit more explicit 😅
English
Alexandru Bucur
2K posts
@nathanjmorton_ Incorrect.
We explicitly discourage RLS (the Postgres feature) in this article, but include an example for completeness since we know it's used quite a bit.
A tenant identifier (tenant_id, user_id, customer_id, etc) does not have to be used in conjunction with RLS.
English
planetscale says RLS is the most scalable multitenant architecture (where tenant = id). thoughts?
PlanetScale@PlanetScale
Postgres has three ways to isolate tenants: - Logical databases - Per-tenant schemas - Tenant ID in a shared schema Counterintuitively, the last is the best way to scale. Read about why in our latest article.
English
English
@Cool_Goose @PlanetScale @samlambert Nope. If you read carefully, it recommends against RLS for row level filtering.
English
Postgres has three ways to isolate tenants:
- Logical databases
- Per-tenant schemas
- Tenant ID in a shared schema
Counterintuitively, the last is the best way to scale. Read about why in our latest article.
English
@alexdaubois Less of a concern in kubernetes in a way :D check pod memory go brr, try to figure it out after (and the pod does restart, so back to slow performance for a bit)
English
@Cool_Goose Lately we exposed per thread memory in FrankenPHP. By watching this metric, it may be possible to warn when a request is leaking. I need to investigate!
English
In traditional PHP-FPM, every request boots a new #PHP environment. Autoloader, framework, DI container... all rebuilt from scratch.
#FrankenPHP worker mode embeds PHP in a Go process. The PHP app boots ONCE and stays in memory.
No fork, no re-init.
English
@alexdaubois I mean profiling should work to identify, i guess "poor man's version" is memory_get_usage/memory_get_peak_usage in php land to at least see if the counter goes up :D i wonder if frankenphp exposes the per request memory somewhere
English
@Cool_Goose That's one of the ideas I have for Ember, I need to understand how this could be possible!
English
it's already standardized on macOS tahoe
the outer window radius (c) is calculated from the window controls radius (a) and controls' distance from the edge of the window (b)
every app that has different controls padding will have different window corner radius, by design
Tailosive Tech@TailosiveTech
Before adding a touch screen to macOS can we at least standardize the corner radius?
English
@saki_statements If only the Netherlands had 130. It's 100 during the day and 120 during the night for 99% of the highways
English
Das Tempolimit in 🇪🇺
🇸🇪 110 km/h
🇳🇴 110 km/h
🇮🇪 120 km/h
🇵🇹 120 km/h
🇪🇸 120 km/h
🇨🇭 120 km/h
🇧🇪 120 km/h
🇳🇱 130 km/h
🇫🇷 130 km/h
🇬🇷 130 km/h
🇭🇷 130 km/h
🇦🇹 130 km/h
🇮🇹 130 km/h
🇩🇰 130 km/h
🇺🇦 130 km/h
🇩🇪 TEMPOLIMIT IST IDEOLOGIE!!! ES GEHT UM UNSERE FREIHEIT!1!1! 😠
Deutsch
@burcs So wild. But I’ll reserve judgement. These open models tend to be benchmaxxed…still promising
English
kimi 2.6 might be the first model that actually replaces claude for me...
and it's open source, wild times!
Matt Silverlock 🐀@elithrar
very excited to see Kimi K2.6 land on Workers AI. benchmarks are benchmarks, but competitive open models drive all models forward.
English
If true, this is a *REALLY* big deal. The first time an open weight model beats the closed labs at SWE-Bench Pro...
Kimi.ai@Kimi_Moonshot
Meet Kimi K2.6: Advancing Open-Source Coding 🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2) What's new: 🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization). 🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D. 🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files. 🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops. 🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop. - K2.6 is now live on kimi.com in chat mode and agent mode. For production-grade coding, pair K2.6 with Kimi Code: kimi.com/code - 🔗 API: platform.moonshot.ai 🔗 Tech blog: kimi.com/blog/kimi-k2-6 🔗 Weights & code: huggingface.co/moonshotai/Kim…
English
@jpschroeder @northflank @Docker @Laravel good scaling, same local environment, batteries included, laravel boost
English
You're starting a greenfield clanker service. What stack are you choosing and why (in the replies)...
English
@adrianmg I guess I am anti industry as technical proficiency and some hands on time is required 😅
English
From time to time I keep reflecting on the idea that, as an industry, we’ve pushed people to see being hands-off as a manager not just as acceptable, but as the mandatory ideal, is still mind-blowing.
It feels deeply wrong to push managers into a state of ignorance and distance, as if being a source of inspiration, judgment, and knowledge were somehow a flaw.
Managers don't need to dictate every direction, or be the most knowledgeable person in the room, but if they are not respected as leaders, how are they supposed to shape teams, build trust, and lead them anywhere meaningful?
English
@devhe4d @CloudflareDev Best env editing for me it's in @northflank since it shows it also as an optional plain. Env file
English
having to copy-paste 20 env vars into a cloudflare project in 2026 is brutal!
@CloudflareDev you are better than this, can we loosen it up a bit? a bulk import option wouldn't hurt anyone :/
English
@gunnarmorling @claudeai It should be an extra one for each thinking level so high has 3
English
Hey @claudeai, why do we have two "Thinking" indicators now? Can we disable that first one, it takes a lot of attention, IMHO.
English
@zeeg Exited to have logs and metrics on sentry 😅 more context mooooore
English
Let it be known: Seer Agent (coming soon) nailed the actual root cause, Devin and Cursor only hit the surface level. Same prompt.
David Cramer@zeeg
for good measure, lets see what Sentry thinks (via Seer) - this isnt apples to apples, but the new agent has access to quite a lot of functionality it found the core problem but also came up with some other wacky opinions (the "actual fix" bit is wrong)
English
Took a bit of effort but you can absolutely get good designs out of Claude Design. Hats off to the team, I dig this.
Theo - t3.gg@theo
Really liking Claude Design so far. Except for the fact that it just wiped out my project after burning 10% of my usage. "The files appear to be gone" 🙃
English
Lenovo changed their keyboard layout??? now I have to undo 6 years of muscle memory thanks....
English
Conclusion, I am an expert prompt engineer. That's all folks.
Andras Bacsai@heyandras
why it could be that opus 4.7 experience is soo different for everyone? i built a very complex new app (rust based) with it, and it was almost one shotted everything.
English