Scott Hanselman

Writing and debugging Linux C++ applications from Visual Studio using the "Windows Subsystem for Linux"

April 3, '17 Comments [19] Posted in Linux | Open Source | Win10
Sponsored By

I've blogged about the "Windows Subsystem for Linux" (also known as "Bash on Ubuntu on Windows") many times before. Response to this Windows feature has been a little funny because folks try to:

  • Minimize it - "Oh, it's just Cygwin." (It's actually not, it's the actual Ubuntu elf binaries running on a layer that abstracts the Linux kernel.)
  • Design it - "So it's a docker container? A VM?" (Again, it's a whole subsystem. It does WAY more than you'd think, and it's FASTer than a VM.)

Here's a simple explanation from Andrew Pardoe:

1. The developer/user uses a bash shell.
2. The bash shell runs on an install of Ubuntu
3. The Ubuntu install runs on a Windows subsystem. This subsystem is designed to support Linux.

It's pretty cool. WSL has, frankly, kept me running Windows because I can run cmd, powershell, OR bash (or zsh or Fish). You can run vim, emacs, tmux, and run Javascript/node.js, Ruby, Python, C/C++, C# & F#, Rust, Go, and more. You can also now run sshd, MySQL, Apache, lighttpd as long as you know that when you close your last console the background services will shut down. Bash on Windows is for developers, not background server apps. And of course, you apt-get your way to glory.

Bash on Windows runs Ubuntu user-mode binaries provided by Canonical. This means the command-line utilities are the same as those that run within a native Ubuntu environment.

I wanted to write a Linux Console app in C++ using Visual Studio in Windows. Why? Why not? I like VS.

Setting up Visual Studio 2017 to compile and debug C++ apps on Linux

Then, from the bash shell make sure you have build-essential, gdb's server, and openssh's server:

$ sudo apt update
$ sudo apt install -y build-essential
$ sudo apt install -y gdbserver
$ sudo apt install -y openssh-server

Then open up /etc/ssh/sshd_config with vi (or nano) like

sudo nano /etc/ssh/sshd_config

and for simplicity's sake, set PasswordAuthentication to yes. Remember that it's not as big a security issue as you'd think as the SSHD daemon closes when your last console does, and because WSL's subsystem has to play well with Windows, it's privy to the Windows Firewall and all its existing rules, plus we're talking localhost also.

Now generate SSH keys and manually start the service:

$ sudo ssh-keygen -A
$ sudo service ssh start

Create a Linux app in Visual Studio (or open a Makefile app):

File | New Project | Cross Platform | Linux

Make sure you know your target (x64, x86, ARM):

Remote GDB Debugger options

In Visual Studio's Cross Platform Connection Manager you can control your SSH connections (and set up ones with private keys, if you like.)

Tools | Options | Cross Platfrom | Connection Manager

Boom. I'm writing C++ for Linux in Visual Studio on Windows...running, compiling and debugging on the local Linux Subsystem

I'm writing C++ in Visual Studio on Windows talking to the local Linux Subsystem

BTW, for those of you, like me, who love your Raspberry Pi tiny Linux computers...this is a great way to write C++ for those little devices as well. There's even a Blink example in File | New Project to start.

Also, for those of you who are very advanced, stop using Mingw-w64 and do cool stuff like compiling gcc 6.3 from source under WSL and having VS use that! I didn't realize that Visual Studio's C++ support lets you choose between a number of C++ compilers including both GCC and Clang.


Sponsor: Thanks to Redgate! Track every change to your database! See who made changes, what they did, & why, with SQL Source Control. Get a full version history in your source control system. See how.

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

Command Line: Using dotnet watch test for continuous testing with .NET Core 1.0 and XUnit.net

March 27, '17 Comments [14] Posted in DotNetCore | Open Source
Sponsored By

I've installed .NET Core 1.0 on my machine. Let's see if I can get a class library and tests running and compiling automatically using only the command line. (Yes, some of you are freaked out by my (and other folks') appreciation of a nice, terse command line. Don't worry. You can do all this with a mouse if you want. I'm just enjoying the CLI.

NOTE: This is considerably updated from the project.json version in 2016.

First, I installed from http://dot.net/core. This should all work on Windows, Mac, or Linux.

C:\> md testexample & cd testexample

C:\testexample> dotnet new sln
Content generation time: 33.0582 ms
The template "Solution File" created successfully.

C:\testexample> dotnet new classlib -n mylibrary -o mylibrary
Content generation time: 40.5442 ms
The template "Class library" created successfully.

C:\testexample> dotnet new xunit -n mytests -o mytests
Content generation time: 87.5115 ms
The template "xUnit Test Project" created successfully.

C:\testexample> dotnet sln add mylibrary\mylibrary.csproj
Project `mylibrary\mylibrary.csproj` added to the solution.

C:\testexample> dotnet sln add mytests\mytests.csproj
Project `mytests\mytests.csproj` added to the solution.

C:\testexample> cd mytests

C:\testexample\mytests> dotnet add reference ..\mylibrary\mylibrary.csproj
Reference `..\mylibrary\mylibrary.csproj` added to the project.

C:\testexample\mytests> cd ..

C:\testexample> dotnet restore
Restoring packages for C:\Users\scott\Desktop\testexample\mytests\mytests.csproj...
Restoring packages for C:\Users\scott\Desktop\testexample\mylibrary\mylibrary.csproj...
Restore completed in 586.73 ms for C:\Users\scott\Desktop\testexample\mylibrary\mylibrary.csproj.
Installing System.Diagnostics.TextWriterTraceListener 4.0.0.
...SNIP...
Installing Microsoft.NET.Test.Sdk 15.0.0.
Installing xunit.runner.visualstudio 2.2.0.
Installing xunit 2.2.0.
Generating MSBuild file C:\Users\scott\Desktop\testexample\mytests\obj\mytests.csproj.nuget.g.props.
Generating MSBuild file C:\Users\scott\Desktop\testexample\mytests\obj\mytests.csproj.nuget.g.targets.
Writing lock file to disk. Path: C:\Users\scott\Desktop\testexample\mytests\obj\project.assets.json
Installed:
16 package(s) to C:\Users\scott\Desktop\testexample\mytests\mytests.csproj

C:\testexample> cd mytests & dotnet test

Build started, please wait...
Build completed.

Test run for C:\testexample\mytests\bin\Debug\netcoreapp1.1\mytests.dll(.NETCoreApp,Version=v1.1)
Microsoft (R) Test Execution Command Line Tool Version 15.0.0.0
Copyright (c) Microsoft Corporation. All rights reserved.

Starting test execution, please wait...
[xUnit.net 00:00:00.5539676] Discovering: mytests
[xUnit.net 00:00:00.6867799] Discovered: mytests
[xUnit.net 00:00:00.7341661] Starting: mytests
[xUnit.net 00:00:00.8691063] Finished: mytests

Total tests: 1. Passed: 1. Failed: 0. Skipped: 0.
Test Run Successful.
Test execution time: 1.8329 Seconds

Of course, I'm testing nothing yet but pretend there's a test in the tests.cs and something it's testing (that's why I added a reference) in the library.cs, OK?

Now I want to have my project build and tests run automatically as I make changes to the code. I can't "dotnet add tool" yet so I'll add this line to my test's project file:

<ItemGroup>
<DotNetCliToolReference Include="Microsoft.DotNet.Watcher.Tools" Version="1.0.0" />
</ItemGroup>

Like this:

Adding <DotNetCliToolReference Include="Microsoft.DotNet.Watcher.Tools" Version="1.0.0" />

Then I just dotnet restore to bring in the tool.

In order to start the tests, I don't write dotnet test, I run "dotnet watch test." The main command is watch, and then WATCH calls TEST. You can also dotnet watch run, etc.

NOTE: There's a color bug using only cmd.exe so on "DOS" you'll see some ANSI chars. That should be fixed in a minor release soon - the PR is in and waiting. On bash or PowerShell things look fin.

In this screenshot, you can see as I make changes to my test and hit save, the DotNetWatcher Tool sees the change and restarts my app, recompiles, and re-runs the tests.

Test Run Successful

All this was done from the command line. I made a solution file, made a library project and a test project, made the test project reference the library, then built and ran the tests. If I could add the tool from the command line I wouldn't have had to manually touch the project file at all.

Again, to be sure, all this is stuff you can (and do) do in Visual Studio manually all the time. But I'll race you anytime. ;)


Sponsor: Check out JetBrains Rider: a new cross-platform .NET IDE. Edit, refactor, test, build and debug ASP.NET, .NET Framework, .NET Core, or Unity applications. Learn more and get access to early builds!

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

dotnet new angular and dotnet new react

February 13, '17 Comments [30] Posted in Javascript | Open Source | VS2017
Sponsored By

I was exploring the "dotnet new" experience last week and how you can extend templates, then today the .NET WebDev blog posted about Steve Sanderson's work around Single Page Apps (SPA). Perfect timing!

image

Since I have Visual Studio 2017 RC and my .NET Core SDK tools are also RC4:

C:\Users\scott\Desktop\fancypants>dotnet --info
.NET Command Line Tools (1.0.0-rc4-004771)

Product Information:
Version: 1.0.0-rc4-004771
Commit SHA-1 hash: 4228198f0e

Runtime Environment:
OS Name: Windows
OS Version: 10.0.15031
OS Platform: Windows
RID: win10-x64
Base Path: C:\Program Files\dotnet\sdk\1.0.0-rc4-004771

I can then do this from the dotnet command line interface (CLI) and install the SPA templates:

dotnet new --install Microsoft.AspNetCore.SpaTemplates::*

The * is the package version so this is getting the latest templates from NuGet. I'm looking forward to using YOUR templates (docs are coming! These are fresh hot bits.)

This command adds new templates to dotnet new. You can see the expanded list here:

Templates                                     Short Name      Language      Tags
------------------------------------------------------------------------------------------
Console Application console [C#], F# Common/Console
Class library classlib [C#], F# Common/Library
Unit Test Project mstest [C#], F# Test/MSTest
xUnit Test Project xunit [C#], F# Test/xUnit
Empty ASP.NET Core Web Application web [C#] Web/Empty
MVC ASP.NET Core Web Application mvc [C#], F# Web/MVC
MVC ASP.NET Core with Angular angular [C#] Web/MVC/SPA
MVC ASP.NET Core with Aurelia aurelia [C#] Web/MVC/SPA
MVC ASP.NET Core with Knockout.js knockout [C#] Web/MVC/SPA
MVC ASP.NET Core with React.js react [C#] Web/MVC/SPA
MVC ASP.NET Core with React.js and Redux reactredux [C#] Web/MVC/SPA
Web API ASP.NET Core Web Application webapi [C#] Web/WebAPI
Solution File sln Solution

See there? Now I've got "dotnet new react" or "dotnet new angular" which is awesome. Now I just "npm install" and "dotnet restore" followed by a "dotnet run" and very quickly I have a great starter point for a SPA application written in ASP.NET Core 1.0 running on .NET Core 1.0. It even includes a dockerfile if I like.

From the template, to help you get started, they've also set up:

  • Client-side navigation. For example, click Counter then Back to return here.
  • Server-side prerendering. For faster initial loading and improved SEO, your Angular 2 app is prerendered on the server. The resulting HTML is then transferred to the browser where a client-side copy of the app takes over. THIS IS HUGE.
  • Webpack dev middleware. In development mode, there's no need to run the webpack build tool. Your client-side resources are dynamically built on demand. Updates are available as soon as you modify any file.
  • Hot module replacement. In development mode, you don't even need to reload the page after making most changes. Within seconds of saving changes to files, your Angular 2 app will be rebuilt and a new instance injected is into the page.
  • Efficient production builds. In production mode, development-time features are disabled, and the webpack build tool produces minified static CSS and JavaScript files.

Go and read about these new SPA templates in depth on the WebDev blog.


Sponsor: Big thanks to Raygun! Join 40,000+ developers who monitor their apps with Raygun. Understand the root cause of errors, crashes and performance issues in your software applications. Installs in minutes, try it today!

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

Trying out "dotnet new" template updates and csproj with VS2017

February 7, '17 Comments [28] Posted in Open Source | VS2017
Sponsored By

I updated my Visual Studio 2017 RC installation today. Here's the release notes. You just run "Visual Studio Installer" if you've already got a version installed and it updates. The updating processes reminds me a little of how Office 365 updates itself. It's not as scary as VS updates of the past. You can download the VS2017 RC at https://www.visualstudio.com and it works side by side with your existing installs. I haven't had any issues yet.

New Templating Engine for .NET Core CLI

It also added/updated a new .NET Core SDK. I am a fan of the command line "dotnet.exe" tooling and I've been pushing for improvements in that experience. A bunch of stuff landed in this update that I've been waiting for. Here's dotnet new:

C:\Users\scott\Desktop\poop> dotnet new
Template Instantiation Commands for .NET Core CLI.

Usage: dotnet new [arguments] [options]

Arguments:
template The template to instantiate.

Options:
-l|--list List templates containing the specified name.
-lang|--language Specifies the language of the template to create
-n|--name The name for the output being created. If no name is specified, the name of the current directory is used.
-o|--output Location to place the generated output.
-h|--help Displays help for this command.
-all|--show-all Shows all templates


Templates Short Name Language Tags
--------------------------------------------------------------------------------------
Console Application console [C#], F# Common/Console
Class library classlib [C#], F# Common/Library
Unit Test Project mstest [C#], F# Test/MSTest
xUnit Test Project xunit [C#], F# Test/xUnit
Empty ASP.NET Core Web Application web [C#] Web/Empty
MVC ASP.NET Core Web Application mvc [C#], F# Web/MVC
Web API ASP.NET Core Web Application webapi [C#] Web/WebAPI
Solution File sln Solution

Examples:
dotnet new mvc --auth None --framework netcoreapp1.0
dotnet new console --framework netcoreapp1.0
dotnet new --help

There is a whole new templating engine now. The code is here https://github.com/dotnet/templating and you can read about how to make your own templates or on the wiki.

I did a "dotnet new xunit" and it made the csproj file and a Unit Test. Here's what's inside the csproj:

<Project Sdk="Microsoft.NET.Sdk">
  <PropertyGroup>
    <OutputType>Exe</OutputType>
    <TargetFramework>netcoreapp1.0</TargetFramework>
  </PropertyGroup>
  <ItemGroup>
    <PackageReference Include="Microsoft.NET.Test.Sdk" Version="15.0.0-preview-20170123-02" />
    <PackageReference Include="xunit" Version="2.2.0-beta5-build3474" />
    <PackageReference Include="xunit.runner.visualstudio" Version="2.2.0-beta5-build1225" />
  </ItemGroup>
</Project>

That's not too bad. Here's a a library with no references:

<Project Sdk="Microsoft.NET.Sdk">
  <PropertyGroup>
    <TargetFramework>netstandard1.4</TargetFramework>
  </PropertyGroup>
</Project>

Note there's no GUIDs in the csproj. Sweet.

Remember also that there was talk that you wouldn't have to edit your csproj manually? Check this out:

C:\Users\scott\Desktop\poop\lib> dotnet add package Newtonsoft.Json
Microsoft (R) Build Engine version 15.1.545.13942
Copyright (C) Microsoft Corporation. All rights reserved.

Writing C:\Users\scott\AppData\Local\Temp\tmpBA1D.tmp
info : Adding PackageReference for package 'Newtonsoft.Json' into project 'C:\Users\scott\Desktop\poop\lib\lib.csproj'.
log : Restoring packages for C:\Users\scott\Desktop\poop\lib\lib.csproj...
info : GET https://api.nuget.org/v3-flatcontainer/newtonsoft.json/index.json
info : OK https://api.nuget.org/v3-flatcontainer/newtonsoft.json/index.json 1209ms
info : GET https://api.nuget.org/v3-flatcontainer/newtonsoft.json/9.0.1/newtonsoft.json.9.0.1.nupkg
info : OK https://api.nuget.org/v3-flatcontainer/newtonsoft.json/9.0.1/newtonsoft.json.9.0.1.nupkg 181ms
info : Package 'Newtonsoft.Json' is compatible with all the specified frameworks in project 'C:\Users\scott\Desktop\poop\lib\lib.csproj'.
info : PackageReference for package 'Newtonsoft.Json' version '9.0.1' added to file 'C:\Users\scott\Desktop\poop\lib\lib.csproj'.

Doing "dotnet add package foo.bar" automatically gets the package from NuGet and adds it to your csproj. Just like doing "Add NuGet Package" (or add reference) in Visual Studio. You don't even have to open or look at your csproj.

I'm going to keep digging into this. We're getting into a nice place where someone could easily make a custom template then "nuget in" that templates then "File | New | Your Company's Template" without needed yeoman, etc.

Please shared your feedback:

Also, be sure to check out the new and growing Docs site at https://docs.microsoft.com/en-us/dotnet


Sponsor: Track every change to your database. See who made changes, what they did, & why, with SQL Source Control. Get a full version history in your source control system. See how with Red Gate's SQL Source Control.

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

Playing with an Onion Omega IoT device to show live Blood Sugar on an OLED screen

December 14, '16 Comments [6] Posted in Hardware | Open Source
Sponsored By

arduino_lb3dg8I've been playing with IoT stuff on my vacation. Today I'm looking at an Onion Omega. This is a US$19 computer that you can program with Python, Node.js, or C/C++. There's a current IndieGogo happening for the Onion Omega2 for $5. That's a $5 Linux computer with Wi-Fi. Realistically you'd want to spend more and get expansion docks, chargers, batteries, etc, but you get the idea. I got the original Omega along with the bluetooth dongle, Arduino compatible base, tiny OLED screen. A ton of stuff to play with for less than $100.

Note that I am not affiliated with Onion at all and I paid for it with my own money, to use for fun.

One of the most striking things about the Onion Omega line is how polished it is. There's lots of tiny Linux Machines that basically drop you at the command line and say "OK, SSH in and here's root." The Onion Omega is far more polished.

Onion Omega has a very polished Web UI

The Omega can do that for you, but if you have Bonjour installed (for zeroconf networking) and can SSH in once to setup Wi-Fi, you're able to access this lovely web-based interface.

Look at all the info about the Omega's memory, networking, device status, and more

This clean, local web server and useful UI makes the Onion Omega extremely useful as a teaching tool. The Particle line of IoT products has a similarly polished web-interfaces, but while the Onion uses a local web server and app, the Particle Photon uses a cloud-based app that bounces down to a local administrative interface on the device. There's arguments for each, but I remain impressed with how easy it was for me to update the firmware on the Omega and get a new experience. Additionally, I made a few mistakes and "bricked" it and was able - just by following some basic instructions - to totally reflash and reset it to the defaults in just about 10 minutes. Impressive docs for an impressive product.

image

Onion Omega based Glucose Display via NightScout

So it's a cool product, but how quickly can I do something trivial, but useful? Well, I have a NightScout open source diabetes management server with an API that lets me see my blood sugar. The resulting JSON looks like this:

[  
{
"_id":"5851b235b8d1fea108df8b",
"sgv":135,
"date":1481748935000,
"dateString":"2016-12-14T20:55:35.000Z",
"trend":4,
"direction":"Flat",
"device":"share2",
"type":"sgv"
}
]

That number under "sgv" (serum glucose value) is 135 mg/dl. That's my blood sugar right now. I could get n values back from the WebAPI and plot a chart, but baby steps. Note also the "direction" for my sugars is "flat." It's not rising nor falling in any major way.

Let's add the OLED Display to the Onion Omega and show my sugars. Since it's an OpenWRT Linux machine, I can just add Python!

opkg update
opkg install python

Some may (and will) argue that for a small IoT system, Linux is totally overkill. Sure, it likely it. But it's also very productive, fun to prototype with, and functional. Were I to go to market for real, I'd likely use something more hardened.

As I said, I could SSH into the machine but since the Web UI is so nice, it includes an HTML-based terminal!

A Terminal built in!

The Onion Omega includes not just libraries for expansions like the OLED Display, but also command-line utilities. This script clears the display, initializes it, and displays some text. The value of that text will come from my yet-to-be-written python script.

#!/bin/sh    

oled-exp -c

VAR=$(python ./sugar_script.py)

oled-exp -i
oled-exp write "$VAR"

Then in my Python script I could print the value that would be returned into VAR and then printed with the oled-exp command line utility.

OR, I can bypass the shell script entirely and use the Python Module for this OLED screen directly and do this. Grab the JSON, clean it up because apparently the json library sucks (?), then display it.

#!/usr/bin/env python                                                                                                        

from OmegaExpansion import oledExp
import urllib
import json

site="https://hanselmansugars.something/api/v1/entries/sgv.json?count=1"
jfile=urllib.urlopen(site)
jsfile=jfile.read()
jsfile=jsfile.replace("\n","")
jsfile=jsfile.replace("/","")
jsfile=jsfile.replace("]","")
jsfile=jsfile.replace("[","")

a=json.loads(jsfile)
sugar=a['sgv']
direction=a['direction']
info="\n" + str(sugar)+" mg/dl and "+direction

oledExp.driverInit()
oledExp.clear()
oledExp.write(info)

Now here's a pic of my live blood sugar on the Onion Omega with the OLED! I could put this to run on a timer and I'm off to the races.

The OLED Screen says "149 mg/dl and Flat"

The next step might be to clean up the output, parse the date better, and perhaps even dynamically generate a sparkline and display the graphic on the small B&W OLED Screen.

Have you used a small Linux IoT device like the Onion Omega?


Sponsor: Do you deploy the same application multiple times for each of your end customers? The team at Octopus have taken the pain out of multi-tenant deployments. Check out their latest 3.4 release

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
Page 1 of 42 in the Open Source category Next Page

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