refactor: retain null values in array

[eaglethrost]

🚥 Resolves ISSUE_ID

🧰 Changes

• This PR is part of the work for the readme ticket RM-9906 Not displaying null values in an array and should get merged first before the changes in the oas repo
• Removes the JSON.parse(stringify) code in favour implementing own function to remove undefined objects so that undefined values don't get converted to nulls, allowing us to separate user added nulls
• In addition, don't remove null values in arrays when stripping objects. I think this is fine todo because from my understanding that code was just to removed the null-converted undefined, please let me know if I'm wrong. Open to feedback on how to approach this.

🧬 QA & Testing

• To the removeUndefinedObjects function, pass in an array with nulls and they should get retained
• Primitive nulls should get retained, undefined should still get filtered

[erunion]

Why is this appearing now?

[kanadgupta]

yeah i guess i'll defer to @erunion on the expected default behavior here — should null values be filtered out of arrays by default? or should we add an add'l option for this?

i do agree that the current JSON.parse(JSON.stringify(obj)) behavior does occasionally yield unexpected results (which we kinda sorta account for):

[erunion]

Yea I just want to make sure there aren't any regressions from how this library behaves currently as we use it in a number of areas within the application.

[eaglethrost]

Yea I just want to make sure there aren't any regressions from how this library behaves currently as we use it in a number of areas within the application.

Yeah that was my concern as well and why I thought about cloning the functionality to the oas repo. To reduce the chance of regression, I agree with Kanad's approach on creating an optional option to preserve nulls. Setting the option on would use the custom remove undefined objects function, instead of stringifying, and not removing nulls in array. This will be set in oas-to-har only for now and other parts of the app that use this package should be unaffected. Do you agree? @erunion

[kanadgupta]

ok after giving it a further think through, i think this is technically fixing unexpected behavior and i'm ok with updating the default behavior as opposed to introducing another flag? based on the README, it looks like we are NOT supposed to remove null values:

i do share @erunion's concerns for regressions though, will defer to him on the final call!

[kanadgupta]

can you add another example that's using the exact same object but with preserveArrayNulls set to false so it illustrates the difference?

[eaglethrost]

Update: Changed current approach to put preserving array nulls under an optional flag to reduce chances of regression. Happy to revert it back to make it the default behaviour as kanad pointed out.

[erunion]

couple minor things but overall im happy with this

[kanadgupta]

one final request otherwise LGTM!

[erunion]

Published to v7.0.0