Fix #148 Add Redis::CommandError handling

[corsonknowles]

Redlock 2.0+ mostly supports object instantiation, except for this edge case.

See:

• Restart redis than get Redis::CommandError: NOSCRIPT No matching script. Please use EVAL. #148
• Error: "No matching script. Please use EVAL." #124

Aside from causing difficulties while upgrading, this limits the use of certain functionality.

I tried to be careful in error handling here that the error class is defined and loaded. This is an extension of the solution offered here:

• Restart redis than get Redis::CommandError: NOSCRIPT No matching script. Please use EVAL. #148 (comment)

Added a spec context for good measure.

[corsonknowles]

@jakephot - since this is essentially your solution re-wrapped, I wonder if you have any comments

[leandromoreira]

that's amazing @corsonknowles thank you! can you add the ostruct to Gemfile or gemspec so we can see the tests running.

[corsonknowles]

that's amazing @corsonknowles thank you! can you add the ostruct to Gemfile or gemspec so we can see the tests running.

I pushed that up and then found

• https://github.com/tagliala/coveralls-ruby-reborn

So switched to address it that way.

[jakephot]

@corsonknowles Looks great! I'm excited to upgrade to 2.x.x finally.

[leandromoreira]

I like the reborn but prefer rebirth 😆 just kidding, it seems we are still pointing to coveralls Failure/Error: require 'coveralls'

[corsonknowles]

I think it's still called that, but I had failed to update the json version, so it was still requiring ostruct transitively from the older version. Updated json version as well and everything's running locally

[leandromoreira]

I think we'll drop support to 2.5 and 2.6 which should be fine, they are EOL anyways.

[leandromoreira]

thank you so much @corsonknowles

[leandromoreira]