is it possible to ingest metadata with a regular H...
# ingestion
p
is it possible to ingest metadata with a regular HTTP request on the DataHub REST API?
c
yes, I have done that. for example: to ingest dataset, use
<http://localhost:8080/datasets?action=ingest>
p
thanks, can i post the sample JSON from the repo to that URL?
c
p
thanks thats really useful
b
yep, thanks Ming. @powerful-egg-69769 would it be helpful if we provided client libraries to ingest? if so, what languages would be best?
p
probably Java & Python would be most useful, as long as they are thin libraries (i.e. not many dependencies)
for now i think the most helpful thing would be some more examples on dataset metadata
👍 2
we can see some super simple ones in the repo; would be interesting to try more complex metadata
b
absolutely. agree the metadata included by default is pretty boring... if you're feel inspired feel free to contribute something a bit more comprehensive 🙂
cc. @mammoth-bear-12532 @gray-shoe-75895 on client lib discussions
p
if you get swagger on your API that is probably enough
since you can auto-generate most client APIs from a swagger endpoint
g
We’re also working on simplifying the install process of the python ingestion libraries - they have a fully typed python class hierarchy which is used to send http requests internally
p
great! if we can set them up without having a dependency on a github repo that would be awesome
right now i think one of the pip install dependencies pulls from a github repo
g
yep I've already asked the maintainer to publish a new version to PyPI