Update method for finding latest ruby version

[kenyonj]

This PR:

• Finds the ruby version by curling the ruby-lang site

Why?

• This allows us to stop maintaining a separate, possibly outdated, version url

[mike-burns]

Finding

[kenyonj]

👍 72dc852

[iwz]

Since we're using rbenv in the laptop script, maybe we could use rbenv install -l to list the available versions and grab the latest version from that list.

[mike-burns]

I shortened this:

curl -s "$ruby_download_site" | sed -ne '/Current stable/{n; /Ruby/s/.*Ruby //; s/<.*//p}'

[kenyonj]

thanks @mike-burns... any thoughts on going this route verse what @iwz suggested? Since this is parsing a webpage it seems like it might have more of a potential to break in the future.

[mike-burns]

Ian's suggestion is probably better.

[kenyonj]

good idea @iwz

[ausmith]

👍 to offloading that to rbenv instead of doing it yourself!

[linkyndy]

I agree, parsing rbenv install -l's versions is the way I see it too. However, it requires some shell ninjas to do the latest version matching

[geoffharcourt]

@kenyonj can you squash these into one commit? I'm soliciting more feedback, but I'd like to get this merged in.

[kenyonj]

@geoffharcourt I've squashed, but I think it would probably good to explore finding the latest version through rbenv instead, however, I use chruby so I haven't experimented with that.

[croaky]

The "it works for RVM and chruby users" argument is appealing. Laptop is coupled to rbenv but it's not a library embedded in people's programs. Many users use Laptop in the following way:

• Fork/clone
• Read the script
• Hack it up to their liking
• Run it

I see the elegance of the rbenv install -l option, but I'm a little more drawn to the current approach's portability in the above scenario. Maybe a little less pretty under the hood but also less work for users?

[iwz]

@croaky but the script installs rbenv. I'm not an rbenv user, so I just removed all the rbenv stuff, installed RVM and my own ruby. no biggie for me.

However, in the default case where no modifications are made to the script, it may be possible to have a race condition where there's a new ruby released that isn't yet available on rbenv, and the script would fail. Unless rbenv is actually doing something similar to this script and just dynamically gets the latest ruby.

[tute]

Where do we serve curl -sSL http://ruby.thoughtbot.com/latest from? Right now it's 2.2.3 instead of 2.2.4.

[croaky]

@tute http://ruby.thoughtbot.com/latest is served from Fastly, backed by a single file on the thoughtbot-ruby S3 bucket. We update it by going to https://tbot.io/s3, signing in, updating the file, and purging the Fastly cache.

I've just updated it to 2.3.0.

[tute]

Thank you!

[croaky]

When I run the script in this branch, I get:

sed: 1: "/Current stable/{n; /Ru ...": bad flag in substitute command: '}'
(23) Failed writing body
failed

I've taken a stab at the rbenv method here:

#429

[croaky]

Went with the rbenv method in 36cd104