Closed smanousopoulos closed 9 years ago
If we accept such an option (say skip_validation
) we must also introduce another state that a dataset can enter during its lifecycle (say draft-with-errors
). Τhus, we must ensure, that this new state doesnt break existing functionality (for example a faulty package doesnt appear at search results etc.).
But, i also think we should try it in a separate branch: it has some valid use cases (basically to assist a user to correct a faulty XML metadata file).
The package HAS to create a valid XML at least so that CSW works...
I agree, i mentioned it at "don't break existing functionality": a package at draft-with-errors
state should be private (to the publisher) and is meant only as a starting point for editing and fixing spotted errors.
A package intentionally created with validation errors (using skip_validation
) should be always positioned at invalid
state.
When creating a package from an XML file that contains errors, it would be useful to create the package with errors, then redirect the user to the edit page with the erroneous values shown (and possibly auto-submit in order to highlight the errors?) in order to correct them