Provide a way like `nvm` to mange .NET SDKs

[doggy8088]

Is your feature request related to a problem? Please describe.

The task of installation does not have a consistent method across different operating systems and platforms. It would be more convenient if managing multiple versions of .NET SDKs, from installation to usage, could be modeled after the way sdkman or nvm does it. I believe this would be a very convenient design and hope the .NET team can consider this suggestion.

Describe the solution you'd like

I hope to provide .NET Version Manager just like nvm:

Examples:

$ dvm use 9
Now using .NET SDK 9.0.301
$ dotnet --version
9.0.301
$ dvm use 8
Now using .NET SDK 8.0.412
$ dotnet --version
8.0.412
$ dvm install 6
Now using .NET SDK 6.0.428
$ dotnet --version
6.0.428

Additional context

N/A

[baronfel]

We definitely agree this would be useful! Today, you can use dnvm to get this kind of behavior, but we have open design specs at dotnet/designs#335 discussing how we can make this kind of feature first-class.

[nagilson]

Interesting that this proposal has a 'use' command 😉 I still think there's value in having a 'use' command.

[baronfel]

The biggest problem with use in our ecosystem is that functionally the only way we have to actually pin the muxer to a specific version is global.json's sdk.version and sdk.rollForward. If the muxer grew the ability to have a specific version forced via an env var or something, I could definitely see a case for dnup use ..... It's just that today the implementation of that gesture would involve either writing a temp global.json that would need cleaning up, or constructing a weird PATH + DOTNET_ROOT so that only the SDK version you mentioned would be a viable candidate for the muxer (also in some kind of temporary way).

[agocke]

https://dnvm.net/

[doggy8088]

@agocke Is dnvm.net supported by Microsoft, or is it a community initiative?

[nagilson]

dnvm.net is @agocke's personal project. It's clear he put a lot of passion into it, and I think it has a lot of valid use cases. He is able to deliver the experience faster, from my perspective, in part due to the lack of it needing to be officially supported by microsoft and deal with all of those caveats 😁 The proposals above are about formalizing a similar idea into an officially supported product. I'll let him add anything else.

[agocke]

@nagilson gave a great answer -- that's my personal project, but it sounds like it fits what you were looking for