You can quote CQL column names to allow any column name that Thrift would allow (suitably encoded for ascii).
For instance, CQL knows that UUIDs are represented as strings like 12345678-1234-5678-1234-567812345678 and will parse them correctly. If you mean the official CompositeType, that should also work fine. If you mean "just using bytes smashed together with a delimiter" then that is supported too, with BytesType data encoded as hex. On Wed, Jul 27, 2011 at 11:48 AM, Anthony Ikeda <anthony.ikeda....@gmail.com> wrote: > For our current project we have decided to use Hector as the client API, > however, with the introduction of CQL I need to understand a few things. > > Firstly, CQL use SQL like constructs. Column names seem to be limited to the > same constraints of SQL (restricted use of delimiters) and yet the strengths > of Cassandra actually lie in the fact that we can delimit column names for > hierarchical use - if anything it was encouraged at the Cassandra SF 2011 > conference. > > Should I be ensuring that I avoid using delimiters such as ':', '-' for > column names now? > > Does CQL Support (Dynamic)Composite column names? Row Keys? > > What other limitations does CQL have that are not present in Hector? > > Thanks, > Anthony > > -- Jonathan Ellis Project Chair, Apache Cassandra co-founder of DataStax, the source for professional Cassandra support http://www.datastax.com
