So my new rule is simple: ✅ Start with frontend + mock data for clarity. ✅ Layer backend as the foundation once UX is real. What about you, do you still start with backend or switched to frontend-first too?

It’s not about ignoring backend best practices. It’s about alignment. Frontend-first development builds trust and feedback early. Backend-first builds stability and scalability later. Together 🤝 that’s where real velocity happens.

That realization changed everything for me. Now I recommend: Frontend first 👉 with mock data. Start with what the client can interact with. Build the flow, the experience, the feeling. Then connect it to real APIs once the direction is validated.

Clients don’t see architecture. They see interfaces. To them, progress means something they can click, move, or visualize. No matter how solid your backend is, if there’s no UI; it feels like nothing’s happening.

I’d spend days setting up clean architecture, repository patterns, domain layers, authentication, caching — basically building a bulletproof system before a single UI was visible. Technically perfect, but here’s the catch…

Frontend or Backend — what do you build first? I used to always start with the backend. As a T-shaped engineer with deep experience in .NET but also strong frontend skills (Angular, React, Vue), my instinct was always to get the architecture right first.

5/ Overall: ✅ Easier CLI tooling ✅ Integrated into SDK ✅ Dev quality-of-life improvements Little things like this show how .NET is evolving toward developer joy. #dotnet10 #DevExperience

4/ Behind the scenes, dnx is part of the .NET CLI itself — so the behavior can evolve with the SDK. Today it runs tools. Tomorrow? Who knows. Maybe dnx becomes your .NET AI-powered butler. 🧠

3/ Sample output 👇 You install and run the tool in a snap — complete with ASCII art vibes: Tool package toolsay@1.0.0 will be downloaded... Proceed? [y/n] (y): y [Cool ASCII Output] Yup, it just works.

2/ What is dnx? Think of it as a CLI sidekick that passes your arguments to dotnet behind the scenes. It simplifies tool usage like this: dnx toolsay "Hello, World!" No more long prefixes. Just run your tool and go. 💨

🧵 Upcoming in .NET 10 SDK: A new era of tool execution begins 1/ Tired of typing dotnet tool exec every time you run a tool? So were we. .NET 10 introduces: 🆕 dnx — a new shortcut script to streamline tool execution. Less typing, more hacking. #dotnet #devtools

So what will be difference between @nuxt_js & @nextjs?

I’m curious: Are you using ULID, UUIDv7, or still stuck on random GUIDs? Let’s talk about what you’re seeing in production. #dotnet #csharp #uuid #ulid #performance #devthread #ai #programmer

This is one of those low-effort, high-impact upgrades. If you’re building anything that writes to a relational DB, especially with clustered indexes, Switching from NewGuid() to CreateVersion7() is a no-brainer.

Why this matters: ✅ Your inserts stay fast ✅ Indexes stay balanced ✅ Your team stays within the BCL ✅ Interop with existing ULID-based systems is still possible

Now .NET 9 gives you UUIDv7. Same benefits as ULID, sortable and time-ordered But fully native via: Guid.CreateVersion7() No dependencies. Just clean, efficient GUIDs.

Developers have been solving this with ULIDs. ULIDs are: •Sortable •Unique •128-bit •More index-friendly due to time order They’ve been the go-to fix — but they require external libraries in .NET.

Here’s why: •Expensive index rebalancing •Poor insert performance •High page fragmentation Your database isn’t happy with random GUIDs.

If you’re still using Guid.NewGuid() as your default ID strategy… It’s time to reconsider. Random UUIDs are fine for uniqueness — but they’re terrible for databases.

Great news from .NET 9: You can now generate time-ordered UUIDs natively using Guid.CreateVersion7(). No third-party libraries. No hacks. Just better database performance, out of the box. 🧵