Is there any plan to support JSON Schema as ingest...
# ingestion
o
Is there any plan to support JSON Schema as ingestion format for Datahub?
b
Actually yes!
It will be a process, however
I mean ingestion can already accept serialized JSON, just the schema is defined using PDL instead of JSON schema
Are you hoping mainly to see the schema of the data more easily?
MCP has an "aspect" field that is serialized aspect JSON corresponding to the aspect PDL
o
I've not used MCP but I have only used MCE so far. By JSON Schema I was referring to this thing - https://json-schema.org/ . Even if it's a subset of MCE/MCP I wouldn't mind being able to push schema information through that format for tools which already generate that. I can skip the MCE generation process. Or for starters, I'd like to at least know what each of them have that the other doesn't.
m
@orange-flag-48535 definitely interested in this thread. Can you describe specifically what would be most useful for you? It seems like "a JSON Schema" spec for the ingestion models so you can produce to them easily.
o
As I understand, JSON Schema lets you provide a list of fields and their data types. We're generating MCEs from one bespoke system right now. But we may not end up using all the MCE features like lineage, tagging for a while. We only need field names and data types to start with. And JSON Schema is good enough for that. If we want to pull from say a 3rd party system tomorrow, they're more likely to export schema in JSON Schema than MCE, that's what I meant.
m
Ah got it. So you’re interested in supporting JSONSchema for the “schemaMetadata” aspect. So that any system can describe its dataset schema using that.
plus1 1
b
Ohhh!