Should configatron.temp ensure that temp_end is called?

[Bob Showalter]

If I write an RSpec example like this:

configatron.temp do
  configatron.foo = "someval"
  expect(something).to be_falsy
end

and the expectation fails, the temp setting "leaks" and can cause other examples to fail unexpectedly. This can be handled by setting configatron.foo to a known value in a before(:example) block, but I wonder if temp should work more like the following:

begin
  configatron.temp_start
  configatron.foo = "someval"
  expect(something).to be_falsy
ensure
  configatron.temp_end
end

[Greg Brockman]

Huh, yeah, I feel like temp should always clean itself up. @markbates, do you agree? Looks like the current behavior has existed forever.

[Mark Bates]

the temp block should always clean after itself. i'm surprised to hear that it's not.

[Mark Bates]

looking at it looks like the temp function does call temp_end, but it should probably have an ensure in there for good measure.

If a user is calling temp_start manually, it's their responsibility to make sure temp_end gets called.

[Greg Brockman]

Yeah, the missing ensure block is what surprised me. Fixing.

[Greg Brockman]

Fixed in 4.3.0: https://github.com/markbates/configatron/blob/master/History.txt#L4