Think And Build

I love Satya, I would sell my dog for Windows!1 Well, I put some reasons against it in the comments and you can ask for more. There’s...a lot. The biggest issues I hear are it's hard and it's a third world OS. Both untrue (finally). Linux Mint Cinnamon comes in Ubuntu or Debian Linux flavors and it works like Windows out of the box, is popular, and free.

And it's increasingly supported. Steam has Proton to run Linux games, you can dual boot for Fortnite, and use WINE to run windows apps ⚠️ If you need to dual boot, keep or install a clean Windows first, and only then install Linux. Because Windows thinks it's god and clears the boot sequence. Still here? How about you... Get and download Linux Mint Cinnamon here (full guide here) Get and download Etcher here Plug a 4GB+ USB into your computer and run Etcher: - Click Select image and select your ISO file - Click Select drive and select your USB stick - Click Flash! Restart the computer, it should boot from USB depending on the boot load order and it'll have the option to press enter on linux mint to boot it, or it'll just auto-boot depending. If it does neither, check here When Linux starts, it's running off USB. Double click the "Install Linux Mint" icon and you're golden!

How to install apps Click Start and type “Software Manager” for a streamlined experience. Or grab stuff online and follow their install instructions for executables. Those end in .appimage and .deb, but you can’t double-click to run it until you right click → Properties → “Allow executing file as program” (true). You can make a start menu entry too.

Using Terminal – uncommon apps or programs for power users might ask you to paste lines in Terminal (equivalent of Command Prompt). Paste them one at a time in order via right click → Paste or Shift+Insert, then press enter after each one. If it starts with “sudo” it’ll run as admin and ask your login password. If it ends with (Y/N) it’s a yes or no question and expects you to type y for yes or n for no. Don’t get overwhelmed, you’ve got this <3

Valuable keyboard shortcuts - Lock is Ctrl+Alt+L now - Run is Alt+F2 now - Crop screenshot to clipboard is Ctrl+Shift+Printscreen Valuable tweaks Desktop icons can’t be dragged right click desktop → Customize → Auto-arrange (off).

For recycle bin right click Desktop → Customize → Desktop Settings → Trash (on).

To display time on a 12-hour clock press Super, type Date & Time → Use 24h clock (off).

For create shortcuts in right-click menu (as "make link") press Super, type Files → Edit → Preferences → Context Menus → Make Links (on). Adding a custom keyboard shortcut “Ctrl+Shift+Esc” to open Process Monitor

I'm used to Task Manager in Windows and Process Monitor is basically that, but it doesn't have a shortcut by default. You can give it one. Start by making a file containing your script, the first line is always needed:

#!/bin/bash gnome-system-mon­itor

Then save it to a stable location with filename ending in .sh Then press Super, type Keyboard → click Add Custom Shortcut, and give it a good name. Browse to the file you made, click add, and click the keyboard binding then press whatever sequence you want.

Adding a custom action “Convert File” to Files Press Super, type Actions → click Download tab → click Convert Files and click the download button to the right of it, then open Terminal, type this, press enter and give your login pass as needed:

sudo apt install ffmpeg imagemagick

I got that from the forum linked by the info button ( i ) for the action. It's a little..."open source" here. Anyway, go to Manage tab and enable it. Now in Files, you can right-click a file to convert it to related formats.

For power users! Disable alt+click to drag windows around Alt+click is used to select multiple lines in programs like VS Code, but window drag interferes with it. Press Super, type Windows → Behavior tab at top → “Special key to move and resize windows” (off)

For power users! Paste paths in filesystem This makes it possible to invoke Run (Alt+F2) and paste any path to open the folder or file, which is a really fast shortcut. Right click start (bottom left) → Customize → Menu tab → “Enable filesystem path entry in search box” (on)

How to run Windows apps in WINE VirtualBox and WinBoat are emulators but WINE isn’t an emulator so it’s faster, when it works. And it tends to work for me. Tutorial to install WINE is here After installing WINE, get WineGUI here

Add WINE programs to start menu searches Use WineGUI to load .exe files to register with WINE. Launchers are scripts displayed like files on your desktop that run on double-click, and we'll use that to run apps through WINE. Right click desktop -> "Create a new launcher here..." -> use any name you want, and this command:

wine "the full path to the windows exe file"

Put it in ~/.local/share/applications to see it in start menu searches.

There are still some rough edges

Files app:

Restricted actions when opened from some apps (can't rename, move files, etc.)

No easy way to copy/paste the address or open an unrestricted version of the Files app from a restricted one, which can get really in the way

Typing defaults to searching the current folder even when Files opens just to save a file (it should default to naming it, then)

Few/no integrations with filetypes e.g. no mp3 tags in properties

Programming has a lot of sub-domains. I've seen a few of them, and I want to describe them and list some topics out because there's just such a breadth of problems to solve out there.

I find a lot of them are in game development because of its union of so many concerns, and desktop or web development have a lot each. You can check out my own projects on github for example.

Input sequencing and mapping — A game like Mortal Kombat works by having you start a sequence of input from a lot of possible choices. As you progress through it, you have a short time to continue the sequence. Hitting the wrong keys can cancel it. Shortcuts can be disabled, like when your character is just hit and needs to recover first. There's a lot to think about in how to design an input system like that, one that might support controllers and remapped input. There are considerations here for all programs, too, about user-defined shortcuts and remappings. Look at how Blender does input comboing to e.g. set values, like x = 36 (enter) as an input sequence that accepts any number of digit keystrokes and even backspaces once it reaches the numeric part of the shortcut.

accessibility in web — This includes voice navigation, color contrast handling, font zoom support, high contrast and reduced animation handling, mobile support, etc. Themeing is wrapped into high contrast, and localization into voice navigation. It can be a complex field!

DSP in images and audio — Digital signal processing (DSP) is the field that backs images or audio. There are so many algorithms out there related to it, and concepts explaining how it works. There are several subsystems to a working audio editor, image or movie editor that are worth exploring. I did both an image editor and simple synth. Some ideas: ribbon brushes, stencil buffer shadow casting, partial fill tool, depth of field stacking and panoramic stitching, colorspace conversions, convolutions and anti-aliasing, bit masking e.g. alpha masks, free distortion or distortion grids, bezier paths and vector tooling, font and image recognition, audio profiling to recognize voice or periods of silence, blend modes, infinite canvases and non-destructive editing, shaders, how JPEG works, audio synthesis and modulation, etc.

interactive fiction — Whether it's just to make a game, or to make the engine itself, IF (interactive fiction) is a whole genre that includes hyperlink and parser games which basically combines storytelling with game mechanics. It can be as simple as static choices to make branching stories, or as complex as stories with variables and entire chunks of RPG logic that plays out in text form. Text-only IF is a popular choice for new programmers who want to explore! See popular engines like Twine or my own

OS systems — File systems, virtual drives, processes, application sandboxing, font rasterization, GPU batching and concepts about security, memory, and redundant error handling. Lots out there!

Common program subsystems — Undo/redo systems, data syncing over network between client/server for one and two-way of 1 or 1:N listeners, state rehydration, generated GUI from state, caching/memoization and batching GPU or others, virtual lists in GUI, GUI repainting and event bubbling...it goes on.

Game subsystems — Plugin architecture and filesystem symmetry for overriding assets (hierarchical or flat), comparing input state to last frame and data interpolation, texture packing and mipmapping. Elements of design for Rogue-likes and rogue-lites, RPGs and MMOs, chat filtering, 3D bumpmap and parallax, recursively destructible terrain in 2D or 3D, Minecraft's infinite file chunk-loading system or its continous noise generation

So many more — The entire domain of coding tools, IoT such as drone control and Pi devices, Anything to do with big data, math parsing and arbitrary integer math, custom syntaxes and script sandboxing, physics AABB bounding box trees and algorithms such as polygon intersection and swept collisions, and curiosities like drawing maps, file structures like RIFF for binary or ZIP syncing, contour map generation, IRL map parsing and route planning -- there are lots to consider! Take inspiration from existing programs, think of all the different types out there when you open one.

Hello, world! Progamming is excitingly unsolved. Anyone can make money (or a difference!) and it's a big mixed bag of problems like instant wealth transfer, data abuse and service denial without law enforcement, but also cultural exposure, free info, and humanitarian software.

What you’re doing matters. You could give more money to your local Bezos, or you could contribute to free projects that create competition and free avenues to avoid other Bezoses. I have opinions on what's better, but this guide has no Jiminy Crickets.

So, what does a programmer need?

Autonomy first and foremost. You must be able to teach yourself new skills

It must be interesting. If it’s just for money, learning will be so much harder

Patience. Code will fail. Debug it, dissect the problem. Solve it. Ask for help

Pragmatism. You want the code to work 100%, but it’s ok if how it works is a little sloppy

How and what to learn?

Learn how to search and find info, and places for it. Do your homework before asking

Learn how to use your IDE software, especially debugging!

Learn basics. There are huge info collections out there (1 2 3), you might like this one

Develop a project or idea. It requires pragmatism, tests search and debug skills, and gives exposure to the software development lifecycle. It’s the one best method. You will learn the advanced stuff as you develop your project

Steps for any project:

Start with a big-picture idea, then define the scope (what features or things it does)

While determining the scope, evaluate feasibility. The project needs to be possible!

Borrow everything. Don’t reinvent the wheel. Search github.com and stackoverflow.com for niche implementations

Organize your feature work into steps that all fit into your scope 

Stay motivated! It’s the big project killer. Mix in exciting work, don’t do it all upfront

Coding habits:

Write first, optimize later. Only write comments or test-cases that you care about

Hi! I've been programming for a decade since college and at Microsoft for ~7 years, and want to share some tips.

Industry & jobs

I applied to 30 jobs a day for 6 months for my first job. In the end, I was picked up by a recruiter that found me on LinkedIn which I wasn't even looking at. They contracted me out for Microsoft and later I was poached by the client who I worked for until the AI bubble this year laid off my team, even managers. My only experience is enterprise, so YMMV. But the industry is young, bubbles like that will continue happening until it settles down. Sidenote: I have strong biases against AI because automation results in job loss and would encourage people not to go that route if at all possible.

Ok, now for...

Debugging & telemetry!

College really didn't teach me this at all. It's also essential that you can at least click run and know how to step in/over functions while debugging. Luckily, I've been coding since 12 years old, but that's...unusual 😅. After decades, here's a recap: With configurations. Most IDEs do this; they have debug and release configurations which you can set up. It can be a format that supports extensive editing, like C# projects in Visual Studio, or an open-ended modular interface like how VS Code lets you make a launch.json file to define configurations, then generates the buttons in the GUI for it. Debugging this way means that you click a button when it's all set up, and it will hit your breakpoints.

By attaching to process. This happens when debugging any program that launches indirectly through another one, like a plugin or accessibility and F12 browser tools. You do this by attaching the debugger to the main process, setting your breakpoints, and waiting for it to load your DLL. I do this in VS and it can be a little finicky because you need to set the debugging options to match your program.

With a sourcemap. A sourcemap takes the compiled, obfuscated or minimized outputs of programs and relates them to lines in the source code so that a breakpoint set in the source code can "find" the instructions in the compiled code. Compiled languages do this behind-the-scenes, but transpiled languages like Typescript needs to link to source and usually has explicit options for managing it. On the desktop side, PDB files are used to debug DLLs, which is valuable for third-party tooling, or to decompile and view it, which some languages like Java or C# can do. Indirectly. I do this surprisingly often, and it's as basic as "visualizing what that for-loop will do" in my mind or recording steps in a text editor. Or it's as complex as searching for exact strings in minified javascript, since they can't be simplified, to figure out where it is in the source code or deal with broken sourcemapping.

Forensics (surprise)! Multi-project products often have logging for unexpected errors that raise signals based on sudden spikes in number, etc. you need to analyze to find the reason. I'd start with checking what triggered the error, then for those logs, what the session IDs or user IDs were. From there, what are those users doing in the datetime range of the error? Sometimes it was just unavailability of upstream services or server balancing issues. Almost always it was a recent change in configuration or code (not when it finished, but when it was pushed to CDN). Cooomplicated.

P.S.: Profiling is its own entire world. I never had to do it, but if you're curious, the inspect tools in most browsers include many types of profilers. There are third-party desktop apps that connect to processes and monitor them, too.