Backend update Dec 12th 2018
Semijoins and antijoins, GROQ reference documentation, and more
GROQ now properly supports semijoins and antijoins, i.e. using ->
and ^
in filters, such as in the following queries:
// Fetch employees in the engineering department *[ _type == "employee" && department->name == "Engineering" ] // Fetch departments with no employees *[ _type == "department" && count( *[ _type == "employee" && department._ref == ^._id ] ) == 0 ]
We have also published complete reference documentation for GROQ.
Enterprise customers are now hosted on a separate, larger search cluster, and have increased attribute limits. SLA agreements now also determine automatic recovery priority in case of outages, as well as the data replication factor across data centers.
Recent bugfixes include:
- Filters following projections were not always applied correctly when dependent on values generated by the projection.
_type
attributes were not validated properly.order()
with multipledesc
expressions did not sort correctly in certain rare cases.- Applying
defined()
and+
directly on join results, e.g.defined(ref->)
, now works properly.
The asset CDN now supports caching of non-image files above 10 MB (up to 5 TB), as well as byte range requests for cached files.