Add `dotnet inspect` command

This is the experience creating a console app.

rich@M1 ~ % dotnet new console -o console     
The template "Console Application" was created successfully.

Processing post-creation actions...
Running 'dotnet restore' on /Users/rich/console/console.csproj...
  Determining projects to restore...
  Restored /Users/rich/console/console.csproj (in 129 ms).
Restore succeeded.

rich@M1 ~ % cat console/Program.cs 
// See https://aka.ms/new-console-template for more information
Console.WriteLine("Hello, World!");

Invariably, the first line of Program.cs will be deleted by the user. That’s unfortunate UX.

Please consider putting/sharing this comment in the template creation output and/or in the project file instead.

Some combination of the following would be great, with a bias to updating the console output only.

rich@M1 ~ % dotnet new console -o console     
The template "Console Application" was created successfully.

See https://aka.ms/new-console-template for more information
on the new .NET 6 templates.

Processing post-creation actions...
Running 'dotnet restore' on /Users/rich/console/console.csproj...
  Determining projects to restore...
  Restored /Users/rich/console/console.csproj (in 129 ms).
Restore succeeded.
rich@M1 ~ % cat console/Program.cs 
Console.WriteLine("Hello, World!");
rich@M1 ~ % cat console/console.csproj 
<Project Sdk="Microsoft.NET.Sdk">

<!--
See https://aka.ms/new-console-template for more information
on the new .NET 6 templates.
-->

  <PropertyGroup>
    <OutputType>Exe</OutputType>
    <TargetFramework>net6.0</TargetFramework>
    <ImplicitUsings>enable</ImplicitUsings>
    <Nullable>enable</Nullable>
  </PropertyGroup>

</Project>

As an aside, there is no mention of async at https://aka.ms/new-console-template. That’s probably a missed opportunity that IMO helps motivate the move to top-level statements.

[richlander]

More ideas:

dotnet inspect foo.dll

It tells you (among other things) which TFM and/or RIDs the binary targets/supports.

dotnet inspect foo.npkg

It tells you (among other things) author, publish date, and dependencies.

This is probably a good project to start and then we can decide if we should integrate it into the SDK or leave it as a .NET tool.

[richlander]

We didn’t design for self-help. We’d have to intentionally design the UX we want and then augment the product to deliver it.

The dotnet list direction looks good. We’d also want a variant that was a complete spew of data to be used for support purposes. This is the analog to docker inspect (which I use a fair bit). docker inspect works the opposite direction. It starts with a huge bag of data and then enables various experiences for chopping it down to what you want. I think that’s the “better way”. It prioritizes richness of data over experience and then transitions the discussion away from fidelity to UX.