Rescue LoadError

[Grafikart]

This fix an issue I had when the system tried to constantize labels

[codecov-io]

Codecov Report

Merging #1500 into master will increase coverage by <.01%. The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1500      +/-   ##
==========================================
+ Coverage   96.83%   96.83%   +<.01%     
==========================================
  Files         205      205              
  Lines       12663    12665       +2     
==========================================
+ Hits        12262    12264       +2     
  Misses        401      401

Impacted Files	Coverage Δ
spec/e2e/node_wrapping_spec.rb	100% <100%> (ø)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update f66ba2e...7c0ae40. Read the comment docs.

[cheerfulstoic]

Thanks for this!

What specifically is the LoadError that you're getting? I'd like to know more about it before we rescue from it.

It would also be good to have a spec which causes this to both explain the issue and to prevent regressions

[Grafikart]

@cheerfulstoic I'm not able to produce the rspec but I can explain how the error is triggered.

Within my rails application :

"FORMATION".constantize
# LoadError: Unable to autoload constant FORMATION, expected /some/random/path/formation.rb to define it

Rails autoloader seems to trigger a LoadError instead of a NameError if it finds a file with the same name. That's why it's hard to create a spec to test this feature.

[cheerfulstoic]

When I try that in my Rails app I get a NameError. Maybe it's a version of Rails thing? What version are you using?

It looks like LoadError is what you get when you try to require something that isn't there:

2.5.1 :003 > require 'aoeundaeound'
Traceback (most recent call last):
        1: from (irb):3
LoadError (cannot load such file -- aoeundaeound)

So I feel like there's something I'm not completely understanding.

Also, I believe that it's ActiveSupport specifically that loads files based on namespace/class conventions and that's a dependency of the neo4j gem, so maybe there is some sort of tool that could be used from the gem to test this.

[Grafikart]

@cheerfulstoic Here are the steps to produce a LoadError on constantize from a fresh Rails app

1. Create a new rails app (without ORM) rails new app -O
2. Create a simple Demo model

app/models/demo.rb

class Demo
end

3. rails c, then "DEMO".constantize will throw a LoadError instead of a NameError

[cheerfulstoic]

OK, thanks. I see how there's an issue where two different class names (Demo and DEMO) map to the same file and then it fails because it can't find DEMO in the file.

It's a bit weird to me because it seems like LoadError comes from Ruby but ActiveSupport is using it for a somewhat separate thing. But it does make me feel better that this is rescuing from a base Ruby exception rather than ActiveSupport. And since the line is so targeted, I don't see a possibility of the other sort of LoadError situation happening erroneously here. So :+1: thanks!

[cheerfulstoic]

I was able to add specs for this. It gave me the opportunity to test for the NameError case as well. Check out this commit: https://github.com/neo4jrb/neo4j/commit/bcae1b4c5fb250fa6b56a82ec4a4fe836b5d56bf

[cheerfulstoic]

Released as 9.2.3. Thanks again!