dgraph-io / dgraph / 6000148643

66%

push

feat(dql): add @unique constraint support in schema for new predicates (#8827)

Partially Fixes #8827
Closes: DGRAPHCORE-206
Docs PR: https://github.com/dgraph-io/dgraph-docs/pull/638

This PR adds support for uniqueness constraint using @unique directive
in DQL schema. This unique directive ensures that all values of the
predicate are different in a Dgraph Cluster. This completes phase 1, and
enables adding a new predicate with unique directive. As part of the
phase 2, we will work on adding support for unique directive for
existing predicates.

## Performance
Live Loader before this change on 21 million dataset took 10m54s whereas
after this change took 11m02s. It did not make any significant different
to non-unique predicates.

## How to Use
You can now specify unique in schema as follows: `email: string @unique
@index(hash) @upsert .`. Now, Dgraph will ensure that no mutation adds a
duplicate for the predicate email.

## Phase 2 [TODO]
- [ ] check if @unique can be added to schema depending upon whether
existing data has any duplicates. If the existing data has any
duplicates, we do not allow adding the @unique directive and return a
query that allows user to identify these UIDs.
- [ ] If index computation is in progress, we should not allow mutations
with predicates for which @unique is set
- [ ] Fix ACL to ensure that we do not end up adding duplicate users
- [ ] Ensure that unique constraint is not violated during Bulk loader

---------

Co-authored-by: Aman Mangal <aman@dgraph.io>

347 of 347 new or added lines in 8 files covered. (100.0%)

58763 of 87560 relevant lines covered (67.11%)

2200726.47 hits per line