Scott Hanselman

Teaching Kids to Code with Minecraft Mods made easy using MakeCode and Code Connection

February 5, '19 Comments [7] Posted in Gaming | Musings
Sponsored By

Back in the day, making a Minecraft mod was...challenging. It was a series of JAR files and Java hacks and deep folder structures. It was possible, but it wasn't fun and it surely wasn't easy. I wanted to revisit things now that Minecraft is easily installed from the Windows Store.

Today, it couldn't be easier to make a Minecraft Mod, so I know what my kids and I are doing tonight!

I headed over to https://minecraft.makecode.com/setup/minecraft-windows10 and followed the instructions. I already have Minecraft installed, so I just had to install the Minecraft Code Connection app. The architecture here is very clean and clever. Basically you turn on cheats in Minecraft and use a local websockets connection between the Code Connection app and Minecraft - you're automating Minecraft from an external application!

Here I'm turning on cheats in a new Miencraft world:

Minecraft Allow Cheats

Then from the Code Connection app, I get a URL for the automation server, then go back to Minecraft, hit "t" and paste it in the URL. Now the two apps are talking to each other.

Connecting Minecraft to MakeCode

I can automate with MakeCode, Scratch, or other editors. I'll do MakeCode.

Make Code is amazing

Then an editor opens. This is the same base open source Make Code editor I used when I was coding for an Adafruit Circuit Playground Express earlier this year.

Now, I'll setup a chat command in Make Code that makes it rain chickens when I type the chat command "chicken." It runs a loop and spawns 100 chickens 10 blocks above my character's head.

Chicken rain

I was really surprised how easy this was. It was maybe 10 mins end to end, which is WAY easier than the Java add-ins I learned about just a few years ago.

Minecraft Chicken Rain

There are a ton of tutorials here, including Chicken Rain. https://minecraft.makecode.com/tutorials

The one I'm most excited to show my kids is the Agent. Your connection to the remote Code Connection app includes an avatar or "agent." Just like Logo (remember that, robot turtles?) you can control your agent and make him build stuff. No more tedious house building for us! Let's for-loop our way to glory and teach dude how to make us a castle!


Sponsor: Get the latest JetBrains Rider with Code Vision, Rename Project refactoring, and the Assembly Explorer. Improved support for C#, VB.NET, F#, TypeScript, and Angular is all included.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

Brainstorming - Creating a small single self-contained executable out of a .NET Core application

February 1, '19 Comments [33] Posted in DotNetCore
Sponsored By

I've been using ILMerge and various hacks to merge/squish executables together for well over 12 years. The .NET community has long toyed with the idea of a single self-contained EXE that would "just work." No need to copy a folder, no need to install anything. Just a single EXE.

While work and thought continues on a CoreCLR Single File EXE solution, there's a nice Rust tool called Warp that creates self-contained single executables. Warp is cross-platform, works on any tech, and is very clever

The Warp Packer app has a slightly complex command line, like this:

.\warp-packer --arch windows-x64 --input_dir bin/Release/netcoreapp2.1/win10-x64/publish --exec myapp.exe --output myapp.exe

Fortunately Hubert Rybak has created a very nice "dotnet-warp" global tool that wraps this all up into a single command, dotnet-warp.

All you have to do is this:

C:\supertestweb> dotnet tool install -g dotnet-warp
C:\supertestweb> dotnet-warp
O Running Publish...
O Running Pack...

In this example, I just took a Razor web app with "dotnet new razor" and then packed it up with this tool using Warp packer. Now I've got a 40 meg self-contained app. I don't need to install anything, it just works.

C:\supertestweb> dir
Directory: C:\supertestweb

Mode LastWriteTime Length Name
---- ------------- ------ ----
d----- 2/6/2019 9:14 AM bin
d----- 2/6/2019 9:14 AM obj
d----- 2/6/2019 9:13 AM Pages
d----- 2/6/2019 9:13 AM Properties
d----- 2/6/2019 9:13 AM wwwroot
-a---- 2/6/2019 9:13 AM 146 appsettings.Development.json
-a---- 2/6/2019 9:13 AM 157 appsettings.json
-a---- 2/6/2019 9:13 AM 767 Program.cs
-a---- 2/6/2019 9:13 AM 2115 Startup.cs
-a---- 2/6/2019 9:13 AM 294 supertestweb.csproj
-a---- 2/6/2019 9:15 AM 40982879 supertestweb.exe

Now here's what it gets interesting. Let's say I have a console app. Hello World, packed with Warp, ends up being about 35 megs. But if I use the "dotnet-warp -l aggressive" the tool will add the Mono ILLinker (tree shaker/trimmer) and shake off all the methods that aren't needed. The resulting single executable? Just 9 megs compressed (20 uncompressed).

C:\squishedapp> dotnet-warp -l aggressive
O Running AddLinkerPackage...
O Running Publish...
O Running Pack...
O Running RemoveLinkerPackage...
C:\squishedapp> dir
Mode LastWriteTime Length Name
---- ------------- ------ ----
d----- 2/6/2019 9:32 AM bin
d----- 2/6/2019 9:32 AM obj
-a---- 2/6/2019 9:31 AM 47 global.json
-a---- 2/6/2019 9:31 AM 193 Program.cs
-a---- 2/6/2019 9:32 AM 178 squishedapp.csproj
-a---- 2/6/2019 9:32 AM 9116643 squishedapp.exe

Here is where you come in!

NOTE: The .NET team has planned to have a "single EXE" supported packing solution built into .NET 3.0. There's a lot of ways to do this. Do you zip it all up with a header/unzipper? Well, that would hit the disk a lot and be messy. Do you "unzip" into memory? Do you merge into a single assembly? Or do you try to AoT (Ahead of Time) compile and do as much work as possible before you merge things? Is a small size more important than speed?

What do you think? How should a built-in feature like this work and what would YOU focus on?


Sponsor: Check out Seq 5 for real-time diagnostics from ASP.NET Core and Serilog, now with faster queries, support for Docker on Linux, and beautiful new dark and light themes.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

Visiting The National Museum of Computing inside Bletchley Park - Can we crack Enigma with Raspberry Pis?

January 29, '19 Comments [4] Posted in Hardware
Sponsored By

image"The National Museum of Computing is a museum in the United Kingdom dedicated to collecting and restoring historic computer systems. The museum is based in rented premises at Bletchley Park in Milton Keynes, Buckinghamshire and opened in 2007" and I was able to visit it today with my buddies Damian and David. It was absolutely brilliant.

I'd encourage you to have a listen to my 2015 podcast with Dr. Sue Black who used social media to raise awareness of the state of Bletchley Park and help return the site to solvency.

The National Museum of Computing is a must-see if you are ever in the UK. It was a short 30ish minute train ride up from London. We spent the whole afternoon there.

There is a rebuild of the Colossus, the the world's first electronic computer. It had a single purpose: to help decipher the Lorenz-encrypted (Tunny) messages between Hitler and his generals during World War II. The Colossus Gallery housing the rebuild of Colossus tells that remarkable story.

A working Bombe machine

The backside of the Bombe

National Computing Museum

Cipher Machine

We saw the Turing-Welchman Bombe machine, an electro-mechanical device used to break Enigma-enciphered messages about enemy military operations during the Second World War. They offer guided tours (recommended as the volunteers have encyclopedic knowledge) and we were able to encrypt a message with the German Enigma (there's a 90 second video I made, here) and decrypt it with the Bombe, which is effectively 12 Enigmas working in parallel, backwards.

Inside the top lid of a working EngimaA working Engima

It's worth noting - this from their website - that the first Bombe, named Victory, started code-breaking on Bletchley Park on 14 March 1940 and by the end of the war almost 1676 female WRNS and 263 male RAF personnel were involved in the deployment of 211 Bombe machines. The museum has a working reconstructed Bombe.

 

I wanted to understand the computing power these systems had then, and now. Check out the website where you can learn about the OctaPi - a Raspberry Pi array of eight Pis working together to brute-force Enigma. You can make your own here!

I hope you enjoy these pics and videos and I hope you one day get to enjoy the history and technology in and around Bletchley Park.


Sponsor: Check out Seq 5 for real-time diagnostics from ASP.NET Core and Serilog, now with faster queries, support for Docker on Linux, and beautiful new dark and light themes.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

NuGet's fancy older sibling FuGet gives you a whole new view of the .NET packaging ecosystem

January 25, '19 Comments [6] Posted in DotNetCore | NuGet
Sponsored By

FuGet diffsI remember when we announced NuGet (almost 10 years ago). Today you can get your NuGet packages (that contain .NET libraries) from Nuget.exe, from within Visual Studio, from the .NET CLI (command line interface), and from Paket. Choice is good!

Most folks are familiar with NuGet.org but have you used FuGet?

FuGet is "pro nuget package browsing!" Creating by the amazing Frank A. Krueger - of whom I am an immense fan - FuGet offers a different view on the NuGet package library. NuGet is a repository of nearly 150,000 open source libraries and the NuGet Gallery does a decent job of letting one browse around. However, https://github.com/praeclarum/FuGetGallery is an alternative web UI with a lot more depth.

FuGet is "advanced mode" for NuGet. It's a package browser combined with an API browser that helps you explore the XML documentation and metadata of a package's assemblies to help you explore and learn. And it's a JOY.

For example, if I look at https://www.fuget.org/packages/Newtonsoft.Json I can also see who depends on the package! https://www.fuget.org/packages/Newtonsoft.Json/dependents Who has taken a public dependency on your package? I can see supported frameworks, namepsaces, as well as internal types. For example, I can explore JToken within Newtonsoft.Json and its embedded docs!

You can even do API diffs across versions! Check out https://www.fuget.org/packages/Serilog/2.8.0-dev-01042/lib/netstandard2.0/diff/2.6.0/ for example. This is an API Diff between 2.8.0-dev-01042 and 2.6.0 for Serilog. This could be useful for users or package maintainers when deciding how big a version bumb is required depending on how much of the API has changed. It also gives you a view (as the downstream consumer) of what's coming at you in pre-release versions!

From Frank's blog:

Have you ever wondered if the library your using has been customized for a certain platform? Have you wondered if it will work on your platform at all?

This doubt is removed by displaying - in full technicolor - all the frameworks that the library supports.

 Supported Frameworks

They’re color coded so you can see at a glance:

  • Green libraries are .NET Standard and will work everywhere
  • Dark blue libraries are platform specific
  • Light blue libraries are for full .NET and Mono only
  • Yellow libraries are old PCLs that we’re all trying to forget

FuGet.org is a fanstatic addition to the .NET ecosystem and I"d encourage you to bookmark it, use it, support it, and get involved!

If you're interesting in stuff like this (and the code that runs stuff like this) also check out Stephen Cleary's useful http://dotnetapis.com/ and it's associated code on GitHub https://github.com/StephenClearyApps/DotNetApis.


Sponsor: Your code is bad, but that’s ok thanks to Sentry’s full stack error monitoring that enables you to track and fix application errors in real time. Stop garbage code from becoming garbage fires.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

How to use Windows 10's built-in OpenSSH to automatically SSH into a remote Linux machine

January 23, '19 Comments [6] Posted in DotNetCore | Hardware | Open Source
Sponsored By

In working on getting Remote debugging with VS Code on Windows to a Raspberry Pi using .NET Core on ARM in my last post, I was looking for optimizations and realized that I was using plink/putty for my SSH tunnel. Putty is one of those tools that we (as developers) often take for granted, but ideally I could do stuff like this without installing yet another tool. Being able to use out of the box tools has a lot of value.

A friend pointed out this part where I'm using plink.exe to ssh into the remote Linux machine to launch the VS Debugger:

"pipeTransport": {
"pipeCwd": "${workspaceFolder}",
"pipeProgram": "${env:ChocolateyInstall}\\bin\\PLINK.EXE",
"pipeArgs": [
"-pw",
"raspberry",
"root@crowpi.lan"
],
"debuggerPath": "/home/pi/vsdbg/vsdbg"
}

I could use Linux/bash that's built into Windows 10 for years now. As you may know, Windows 10 can run many Linuxes out of the box. If I have a Linux distro configured, I can call Linux commands locally from CMD or PowerShell. For example, here you see I have three Linuxes and one is the default. I can call "wsl" and any command line is passed in.

C:\Users\scott> wslconfig /l
Windows Subsystem for Linux Distributions:
Ubuntu-18.04 (Default)
WLinux
Debian
C:\Users\scott> wsl ls ~/
forablog forablog.2 forablog.2.save forablog.pub myopenaps notreal notreal.pub test.txt

So theoretically I could "wsl ssh" and use that Linux's ssh, but again, requires setup and it's a little silly. Windows 10 now supports OpenSSL already!

Open an admin PowerShell to see if you have it installed. Here I have the client software installed but not the server.

PS> Get-WindowsCapability -Online | ? Name -like 'OpenSSH*'

Name : OpenSSH.Client~~~~0.0.1.0
State : Installed

Name : OpenSSH.Server~~~~0.0.1.0
State : NotPresent

You can then add the client (or server) with this one-time command:

Add-WindowsCapability -Online -Name OpenSSH.Client~~~~0.0.1.0

You'll get all the standard OpenSSH stuff that one would want.

OpenSSL tools on Windows

Let's say now that I want to be able to ssh (shoosh!) into a remote Linux machine using PGP keys rather than with a password. It's much more convenient and secure. I'll be ssh'ing with my Windows SSH into a remote Linux machine. You can see where ssh is installed:

C:\Users\scott>where ssh
C:\Windows\System32\OpenSSH\ssh.exe

Level set - What are we doing and what are we trying to accomplish?

I want to be able to type "ssh pi@crowpi" from my Windows machine and automatically be logged in.

I will

  • Make a key on my Window machine. The FROM. I want to ssh FROM here TO the Linux machine.
  • Tell the Linux machine (by transferring it over) about the public piece of my key and add it to a specific user's allowed_keys.
  • PROFIT

Here's what I did. Note you can do this is several ways. You can gen the key on the Linux side and scp it over, you can use a custom key and give it a filename, you can use a password as you like. Just get the essence right.

Below, note that when the command line is C:\ I'm on Windows and when it's $ I'm on the remote Linux machine/Raspberry Pi.

  • gen the key on Windows with ssh-keygen
  • I ssh'ed over to Linux and note I'm prompted for a password, as expected.
  • I "ls" to see that I have a .ssh/ folder. Cool. You can see authorized_keys is in there, you may or may no have this file or folder. Make the ~/.ssh folder if you don't.
  • Exit out. I'm in Windows now.
  • Look closely here. I'm "scott" on Windows so my public key is in c:\users\scott\.ssh\id_rsa.pub. Yours could be in a file you named earlier, be conscious.
    • I'm type'ing (cat on Linux is type on Windows) that text file out and piping it into SSH where I login that remote machine with the user pi and I then cat (on the Linux side now) and append >> that text to the .ssh/authorized_keys folder. The ~ folder is implied but could be added if you like.
  • Now when I ssh pi@crowpi I should NOT be prompted for a password.

Here's the whole thing.

C:\Users\scott\Desktop> ssh-keygen
Generating public/private rsa key pair.
Enter file in which to save the key (C:\Users\scott/.ssh/id_rsa):
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
Your identification has been saved in C:\Users\scott/.ssh/id_rsa.
Your public key has been saved in C:\Users\scott/.ssh/id_rsa.pub.
The key fingerprint is:
SHA256:x2vJHHXwosSSzLHQWziyx4II+scott@IRONHEART
The key's randomart image is:
+---[RSA 2048]----+
| . .... . |
|..+. .=+=. o |
| .. |
+----[SHA256]-----+
C:\Users\scott\Desktop> ssh pi@crowpi
pi@crowpi's password:
Linux crowpi 2018 armv7l

pi@crowpi:~ $ ls .ssh/
authorized_keys id_rsa id_rsa.pub known_hosts
pi@crowpi:~ $ exit
logout
Connection to crowpi closed.
C:\Users\scott\Desktop> type C:\Users\scott\.ssh\id_rsa.pub | ssh pi@crowpi 'cat >> .ssh/authorized_keys'
pi@crowpi's password:
C:\Users\scott\Desktop> ssh pi@crowpi
pi@crowpi: ~ $

Fab. At this point I could go BACK to my Windows' Visual Studio Code launch.json and simplify it to NOT use Plink/Putty and just use ssh and the ssh key management that's included with Windows.

"pipeTransport": {
"pipeCwd": "${workspaceFolder}",
"pipeProgram": "ssh",
"pipeArgs": [
"pi@crowpi.lan"
],
"debuggerPath": "/home/pi/vsdbg/vsdbg"
}

Cool!

NOTE: In my previous blog post some folks noted I am logging in as "root." That's an artifact of the way that .NET Core is accessing the GPIO pins. That won't be like that forever.

Thoughts? I hope this helps someone.


Sponsor: Your code is bad, but that’s ok thanks to Sentry’s full stack error monitoring that enables you to track and fix application errors in real time. Stop garbage code from becoming garbage fires.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

Disclaimer: The opinions expressed herein are my own personal opinions and do not represent my employer's view in any way.