OpenAPI schema mismatch with actual database data

[fopina]

Currently the OpenAPI schema is generated with many validations derived from model definitions.

However, some of these are not actually enforced when adding data via the UI or, even worse, from parsers, that actually skip many of them (all?).

I have been maintaining and actively using a python package based on OpenAPI generator and I'm tracking some of these differences in fopina/defectdojo-api-generated#39

For my library in particular, I've been disabling most of the validations in a pre-processing step of the schema but should this be addressed for any OpenAPI consumer?

Given a validation in the schema, it is expected that data coming from the API should already be valid, not invalid...

As a different note, can I open a PR to docs to add my library to the list or what is the process to suggest that? I'd be happy to have more users providing feedback 😄

[valentijnscholten]

Ideally we minimize mismatches, but I am not sure I understand the question or proposal. What exactly are you proposing?

[fopina]

Sorry, I guess I could have proposed something immediately, I was trying to understand if it was seen as an issue first or if it was just me.

I guess my go-to solution would be enforce the validations on the parsers so data is not invalid (parsers to deal with invalid data however they should).
This would actually provide clean, normalized, fields.

Or, the second option, just drop all validations entirely from the schema (maybe even from model definitions), so it becomes consistent.