opendataservices / compile-to-json-schema Goto Github PK

Can we use orderedDict to preserve ordering as much as possible when resolving references

adds additionalProperties = false at every level.

Option off be default, obvs.

https://github.com/openownership/spreadsheet-to-schema

That library does some of the same things (eg codelist files) so I am interested in possible links?

as noted in #4.

Can pass directory to tool, this contains the code lists.

For any closed codelists, take codes from CSV files and put into ENUM's.

It's unclear whether the tool is compatible with newer, or multiple, versions of JSON Schema. This is important because the behaviour of $ref is different between drafts 04–07, and I can't tell whether the tool respects this automatically or not.

https://json-schema.org/understanding-json-schema/reference/combining.html

This may be a use case:

In 360 DEI, we have one schema block repeated 3 times. So it's a ref because that's sensible. But in this block there is a codelist.

Do we want to be able to produce a file with the codelist options included as a ENUM, but with the block still in one definition and with ref's to it? Such a file would be more human readable

Currently we can only produce a file with the same schema repeated 3 times which isn't very clear to humans

From Open-Telecoms-Data/open-fibre-data-standard#57 (comment)

It's not for BODS, it's for JSON Schema

When references are resolved, the 'definitions' section of the schema becomes effectively redundant.

It may be good to by default remove this, with an option to retain it if there is some use-case which requires it.