> > It's also worth pointing out that making this kind of filtering performant > is _not_ a requirement (okay, using Spark below is too slow ;)). What I am > asking for here is a tooling feature, not something I would use for > production loads. Personally, I like how CQL is somewhat locked down for > usage patterns that performs well (that is only filtering based on > indexes). That's the reason why I proposed this sort of thing to be added > to cqlsh. > > That's why we have ALLOW FILTERING. Now, there is many cases where query are not supported even with ALLOW FILTERING due to silly implementation reason but that will solved soon (see CASSANDRA-6377 and CASSANDRA-4987).
So we will most likely support your use case server side soon, protected behind the ALLOW FILTERING flag. But we definitively don't want to do this only in cqlsh because having some type of query being handled behind the scene by cqlsh would be 1) confusing for users 2) probably messy code-wise for cqlsh and 3) rather unexpected for a tool that is just supposed to be a relatively simple shell for cql, not some complex tooling. -- Sylvain
