Hi there,
I am increasingly thinking that a version of the glossary based on the V2
APIs is the sensible way forward, instead of the current fix for 1186,
which uses the V1 taxonomy API and is knitted to the V1 entity APIs and
accesses the graph directly, and manages relationships outside the type
system.
I am looking into a V2 of the Glossary. I am thinking...
1) We could define 3 glossary types , glossarytaxonomy, glossaryterm and
glossarycategory
2) This would mean that terms would be entities (as per recent discussion
on the dev list) and have a 1 to 1 correspondence with graph vertices.
3) We would need Atlas to manage bidirectional links as raised in
ATLAS-1459. It is possible to model parent and children relationships
without it - but this gives the API more integrity .
4) We define these types in a json model file (similar to the existing
bridges), then generate POJO API project, the REST API and bridge code (to
allow import / export for glossary items). A generic mechanism to be able
to do this would make using new types more accessible for Atlas users. I
think the POJO and rest APIs would be more intuitive if they included the
type names. With Rest end points starting something like
/V2/namedentities/glossarytaxonomy. There needs to be places in this
generated code to customise; for example to be able to add extra API
endpoints and constraints - we would need this for the glossary. I have
raised this in Jira ATLAS-1462.
5) If a term is available as a tag then there would be a classification
based on the term that would be used to classify actual assets (e.g a Hive
column). We would then not have the 2 classifications associated with
terms. We would need to think of how we tie the term to its tag. It
suggest implementing classifications as entities as well - to give them
guids and to use a standard Atlas relationships.
How does this sound?
all the best, David.
----- Forwarded by David Radley/UK/IBM on 13/01/2017 11:47 -----
From: David Radley/UK/IBM
To: dev@atlas.incubator.apache.org
Date: 05/01/2017 18:37
Subject: Glossary direction and ATLAS-1186
hi,
I wonder what the community is thinking about the direction we want to
take around Apache Atlas Glossary. I have raised a number of Jiras with a
series of proposals with the idea to create a more mature glossary. The
first one (ATLAS-1186) is in review, all the issues are addressed, but
nothing has happened for 3 weeks.
I am pretty new to this process, so I appreciate any advice or suggestions
on how this is likely to proceed, and what I can do to help,
many thanks, David.
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
