This RV is ridiculous

[DannyBen]

I am now some 24 hours into my "lets try rv" experience, and I feel compelled to share.

I am a long time Ruby developer, with many public ruby gems, and I have my own "mini ruby ecosystem" of my own gem chains and workflows. I have been using RVM in the past (loved it), and later switched to rbenv, which provided a stable ruby installation experience for me.

Now, I feel like rv is the most exciting thing that happened in ruby-land since we started applauding Sinatra.

The snappiness of doing things with rv is just insulting. Everything takes so little time, with so few messages printed, that it seems unrealistic. Even Ruby itself feels significantly faster. All my Ruby-based CLIs run near-instantly, almost as if they were compiled binaries.

Before rv, installing a new ruby version was by far the most intensive operation in my setup. Now, it moved to the opposite side of the scale. It is as snappy as "echo install ruby please".

One thing I love about the Ruby ecosystem, is that it was able to keep a sane, single dependency manager up until now (contrary to the node ecosystem for instance), so I was very reluctant to try rv, concerned it will fragment the Ruby ecosystem, but seeing the magnitude of improvements it brings - I sure hope it evolves fast enough to support all the tasks it set its sights on, and that it gains traction quickly.

For now - I am using rv mostly for installing my single Ruby version, as I never needed more than one version installed, and still using bundle to manage my dependencies (although rv ci seems to be operational and blazing fast as well).

To the developers I say, for whatever its worth: This Ruby developer is behind you.

I am happy to help wherever I can, although rust is outside the scope of my skills, I am happy to test new features, edge cases and usability/UX design.

[kaspth]

Now, I feel like rv is the most exciting thing that happened in ruby-land since we started applauding Sinatra.

Thank you very much for the kind words!

Even Ruby itself feels significantly faster. All my Ruby-based CLIs run near-instantly, almost as if they were compiled binaries.

I don't see anything that makes this true, but it's funny because it feels that way to me too.

One thing I love about the Ruby ecosystem, is that it was able to keep a sane, single dependency manager up until now (contrary to the node ecosystem for instance), so I was very reluctant to try rv, concerned it will fragment the Ruby ecosystem, but seeing the magnitude of improvements it brings - I sure hope it evolves fast enough to support all the tasks it set its sights on, and that it gains traction quickly.

Ironically the point was to experiment with a bunch of new ideas in a new Rust-based codebase to see what things are possible, and then bring whatever fits back to that one single dependency manager. But well, things happened :)

For now - I am using rv mostly for installing my single Ruby version, as I never needed more than one version installed, and still using bundle to manage my dependencies (although rv ci seems to be operational and blazing fast as well).

Yeah, and just using rv purely for Ruby installs is one of the intended use cases, so that should always work. That said please do try out rv ci and let us know if there's things we could improve :)

I am happy to help wherever I can, although rust is outside the scope of my skills, I am happy to test new features, edge cases and usability/UX design.

The next thing we're working on is rv tool install, hopefully we'll have a release version where we're looking for testing soon.

[DannyBen]

Perhaps one reason I am not yet using rv ci is that I am not 100% sure what it does, and how it compares to bundle commands.
It is comparable to bundle install or to bundle update?

A typical use case for me is to bundle update in a gem repository (with Gemfile and *.gemspec) and then to build and install the gem locally.

I usually run

bundle update --all && bundle outdated
gem build ...
gem install --local ...

[DannyBen]

it's mostly meant for CI to install from a clean slate

This could be useful, but for this we need an official Github Action to replace ruby/setup-ruby no?

    - name: Setup Ruby
      uses: ruby/setup-ruby@v1
      with: 
        ruby-version: '${{ matrix.ruby }}'
        bundler-cache: true

EDIT: I see it is somewhat requested in #167

[kaspth]

Yep, we're working on that action right now :)

[indirect]

To me, the simplest explanation of rv ci is that it is the same as bundle install --frozen. You can install the gems that are already locked, but you can't make changes to the Gemfile that would mean the Gemfile.lock needs to be updated.

[eirvandelden]

Just one more question on rv ci:
I currently use binstubs with bin/ at the start of my PATH so I always try to run the binstubs version first.
However, when I switch to using rv ci, I should remove that and stop using binstubs? Because rv already fixed my PATH?

[indirect]

@eirvandelden this is one of those things that is unfortunately confusing even before adding rv, and rv being half-finished means it's just a little bit worse on top, sorry about that.

Today, rv only edits your PATH for ruby itself (and the commands built in to every version of ruby, like gem). It does not edit your path to provide commands from gems in your Gemfile.

The rv ci command is exclusively a faster replacement for bundle install --frozen. Everything else continues to run through bundle exec or bin/rails or whatever the already-existing project setup is.

Our next focus for rv is adding direct support for projects. Reading the Gemfile, adding a project config file, updating the lockfile when you edit the Gemfile, and running per-project tools. Keep an eye on rv's GitHub releases, or the Spinel blog to learn