Sunday, November 15, 2026

An OS which Runs All Apps of All Platforms

Current platforms with the most users
  • Mobile: Android, iOS
  • Desktop: Windows, macOS, Linux
  • Web: Browsers
It's not possible out-of-the-box to run iOS or macOS in virtual machines such as VirtualBox, VMware; as these OSes check the hardware whether it's Apple's, and also check the licence (maybe even to match hardware id).

For Android, it's not convenient to run VMs on it 'cause each app on Android has isolated file system and the VM can't access the global file system.

The remaining choices are: Windows, Linux.

A single taskbar for apps

All apps (not Apple's) can run on Windows this way:
  • Windows apps: Native environment
  • Android apps: Emulator & multi-display (Android 10+)
  • Linux apps: WSL2g
All apps (not Apple's) can run on Linux similarly:
  • Windows apps: VirtualBox (with licence) & one virtual display per app
  • Android apps: Emulator & multi-display (Android 10+)
  • Linux apps: Native environment

Friday, July 10, 2026

The Job Scope Change of Software Developer during this AI Era

The software development process has been being with these steps either without (before 2022, ChatGPT) or with AI agents:
  • Technical design
  • Guiding (if with subcoordinator)
  • Coding
  • Reviewing
Technical design is by the aim points of a project, or the overall view to do. This has not been changed in this AI era if specific features are needed instead of letting AI 'invent' things out.

Guiding is needed in AI era, AI agent needs you to tell what to do.

Coding: This part can be excluded, if having engineer-level AI agents, not the intern-level free AIs. But for productivity, sometimes, dev codes manually better if connecting multiple things the complex way or too specific technical details that is when telling AI in natural language is taking even more time than typing code.

Reviewing is still a much to ensure product quality from inside out, not just testing UI (outside) only.

Wednesday, July 1, 2026

Vibe/Tab/Code: And the 2 Confusing Branches of Vibe Coding

From the start of AI race (ChatGPT 2022), people started to have the concept of vibe coding, or just vibing. This trend is growing, but there's a confusion there.

Everybody used to make programmes (apps) manually and it's called: Coding (verb: code).

And with the AI model for coding available, better productivity is now achieved with assisted coding with Tab key to fill suggestions after caret, this is called: Tabbing (verb: tab).

Vibe coding (Vibing) is the highest point where a confusion is: Review the generated code or no? This splits into 2 branches:
  • Software Industry: Vibe coding with serious reviewing
    • eg. Google Antigravity IDE
  • Social People: Vibe coding with NO reviewing
    • eg. Google Antigravity 2
All can make apps without reviewing, but that's less quality
  • With reviewing & testing: Quality from inside, and outside
    • Full quality
  • Without reviewing, testing only on UI: Quality from outside only
    • Half quality

Tuesday, June 30, 2026

TensorFlow is More Preferred for Those Favour Neuroscience

There are 2 most popular machine learning (ML) libraries called TensorFlow and PyTorch. But for those who favour neuroscience, PyTorch is somehow wrong.
The activation function is around the axon hillock area which is inside the neuron, inside neuron layer. TensorFlow has the activation inside layer, that is matching with neuroscience. However, PyTorch has activation outside of layer which is not complying with neuroscience.

After axon hillock, the value flow is kinda single value, to every axon ternimals, the value should be the same. Not completely the same, because it's biology not digital value.

Saturday, June 27, 2026

The Mouse Scroll in PC vs Mac: A Psychology View

Everybody knows the mouse scroll directions of PC (Windows) and Mac (macOS) are different. There's a psychology behind of both.

Windows
  • When user thinks of something below
    • Scroll down (below)
  • When user thinks of something above
    • Scroll up (above)
macOS
  • User thinks of the UI as a paper page
  • Scroll up to push it up
  • Scroll down to push it down

Google Antigravity IDE and the Extension Store

By default Google Antigravity uses the Open VSX extension store, not everything is there. In case a dev needs extensions from VS Code. The belowing shows how.

Google AI Mode: 
Point Antigravity to the VS Code Marketplace
You can permanently change Antigravity’s extension gallery settings to pull from Microsoft's marketplace instead of Open VSX. [1]
  1. Open Antigravity Settings using Ctrl + , (or Cmd + , on Mac). [1]
  2. Search for Marketplace or navigate to Editor settings. [1]
  3. Locate the inputs for the marketplace URLs and update them to the following:
    • Marketplace Item URL: https://marketplace.visualstudio.com/items
    • Marketplace Gallery URL: https://marketplace.visualstudio.com/_apis/public/gallery [1, 2, 3]
  4. Restart Antigravity IDE. You will now see the JetBrains.kotlin-server extension directly when you search in the extensions tab. [1, 2, 3, 4]

Friday, June 26, 2026

Frameworks for Mobile and Desktop: NativeScript and Electron Similarity

NativeScript
NativeScript has V8 engine inside, but it uses only the JS part, when UI is still converted to native UI and it doesn't use a web-based UI as in Ionic/Cordova/Capacitor.
  • The same as Electron
    • ✅JS: From V8 engine
  • Different from Electron
    • 🔕Chromium UI: XML/CSS are converted to native, no Chromium
Electron
For the desktop side, Electron is also using V8 engine with web-based UI, however, web-based UI on desktop environment works perfectly smooth and no UX issue.
  • The same as NativeScript
    • ✅JS: From V8 engine
  • Different from NativeScript
    • ✅Chromium UI: Using Chromium UI directly

Thursday, June 25, 2026

Cross-platform Mobile App Frameworks and Local Database

The Frameworks
React Native is going backward in time when mixing code & XML for UI when the web world has separated JS and HTML files long ago. Other JS-based frameworks are described below.

JS/TS-based frameworks for mobile apps:
  • NativeScript: Good choice, handled by top open-source JS team OpenJS Foundation
  • React Native: XML mixed in JS is not the way, work-around: split XML to separate files
  • Titanium: Handled by TiDev a less popular organization, longevity issue.
  • Tabris.js: Handled by a company in Germany, also not popular for longevity.
Non-JS/TS frameworks (not convenient, JS/TS is about community and flexibility of coding)
  • Swift UI: Is for iOS only (mostly)
  • Kotlin MP: Is just Java-equivalent, not flexible JS/TS
  • Flutter (Dart), MAUI (C#)
Webview-based frameworks with UX issues (eg. not smooth)
  • Ionic, Cordova, Capacitor, Quasar
The Databases
Mobile apps have gone blossoming for years, more advanced mobile apps need local database similar to IndexedDB in browser for web apps.

Good local databases for mobile apps
  • Couchbase: NoSQL, supported in NativeScript as Java interop; top NoSQL beside MongoDb.
  • SQLite: SQL, supported in many frameworks
Note about Couchbase for NativeScript

Wednesday, June 24, 2026

Google vs Bing: A Small Feature can Make the Win

We have entered AI era (since ChatGPT 2022), and the AI race is here. AIs need data, live data, thus Google and Bing have advantage, with their own existing search engine data.

They're both big companies with big thinkers, but from my opinions, a small UI feature can make the win for Google, and even more win for Google as it has already the top visitors to google.com.

The winner Google is with what users want, convenience to access both search engine and AI.

Google UI:


Bing UI:



The Available Free AI APIs and Their Future

The Internet is with 2 major eras until now: The search engine era, and now is the AI era.

In the search engine era, there were many search engines and concluded with Google and Bing as the 2 major search engines. Search API has never been free.

The landscape of AI era is a bit different, there are free AI available, but why are they free?
  • Search engine API
    • Search engines take the data from the Internet (single source)
    • The queries by users weren't much for using
  • AI API
    • AIs take data from the Internet and users (two sources)
    • The conversations are useful for training AI
That's a difference to be free, and free AI APIs always take your data for training. Unless it's local AI, running on your device but for this case it's not about API.

An OS which Runs All Apps of All Platforms

Current platforms with the most users Mobile: Android , iOS Desktop: Windows , macOS, Linux Web: Browsers It's not possible out-of-the-b...