I've studied the source code and I don't believe that statement is true. I've chatted with several long time users of Cassandra and there's things CQL3 doesn't support.
Like I've said before. Thrift and CQL3 compliment each other. I totally understand some committers don't want the overhead due to time and resource limitations. On more than one occassion, people have offered to help and work on thrift, but were rejected. There's logs in jira. For the record, it's great that CQL was created to make life easier for new users. But here's the thing that annoys me. There's users that just want to save and query data, but there's people out there like me that are building tools for Cassandra. For tool builders, having object API like thrift is invaluable. If we look at relational databases, we see many of them have 2 separate API for that reason. Microsoft SqlServer has SQL and object API. Having both makes it easier to build tools. It's a shame to ignore all the lessons RDBMS can teach us and suffer NIH syndrome. I've built several data modeling tools over the years including ORM's. We built our own data modeling tool for the temporal database I built on Cassandra, so this isn't just some hypothetical complaint. This is from many years of first hand experience. I understand my needs often don't and won't line up with what's in Cassandra's roadmap. But that's the great thing about open source. Should thrift go away permanently I'll just fork Cassandra and do my own thing. On Wed, Jan 21, 2015 at 8:53 AM, Sylvain Lebresne <[email protected]> wrote: > On Wed, Jan 21, 2015 at 3:46 AM, Peter Lin <[email protected]> wrote: > >> >> I don't understand why people [...] pretend it supports 100% of the use >> cases. >> > > Have you consider the possibly that it's actually true and you're just > wrong by lack of knowledge? > > -- > Sylvain >
