Regression in support of sort in queries [DATACASS-842]

[spring-projects-issues]

Luc Boutier opened DATACASS-842 and commented

When migrating a project from cassandra 2.x to 3.x I encountered an issue in the ordering of sort parameters.
The clustering key of my object contains (in order) fields repairOrderId and revisionId and I was using the following query

val query = query(where("partitionKey").is(partitionKey)).sort(by(DESC, "repairOrderId", "revisionId"))
return cassandraTemplate.selectOne(query, RepairOrderByBasket::class.java)

It seems that somehow the sort is mixing up and creates a statement with 

ORDER BY revisionId DESC, repairOrderId DESC

that cassandra does not support (it requires the same order as clustering keys so 

ORDER BY repairOrderId DESC, revisionId DESC

In my particular case it was actually just better to use a clustering ordering on the table definition rather than at the query level but I guess this can be an issue in some other context

---

No further details from DATACASS-842

[mp911de]

Thanks for the report @lucboutier. That's caused by collecting sort orders in a Map which messes up the actual order of sort-by properties.

[mp911de]

That's fixed now.

[lucboutier]

Amazing! Thanks a lot for the great work!