A current implementation in a language you are familiar with is always a good place to start. However, be aware of language differences -- python for example doesn't have a float datatype, only double, so it has to track that outside of the language typing system.
There have been several discussions on the dev list that might help too -- http://search-hadoop.com works great for searching for info in past discussions. In general, the two most tricky Avro features are Unions and Schema resolution/evolution. These are in many ways the most useful and powerful features of Avro. Thinking about how to approach those ahead of time will save you time in the log run. Each language needs to figure out how best to represent Unions. Most don't have language level type-safe ways of dealing with them, but some do: http://code.google.com/p/avro-scala-compiler-plugin/ Be aware that we have discovered recently that not all languages do schema resolution of complicated unions consistently. I hope that helps! -Scott On 1/21/11 4:39 PM, "R. Tyler Croy" <[email protected]> wrote: >I've been experimenting with Ada in my free time, stop laughing, I'm >serious! >Anyways, I'm thinking about implementing Ada bindings for Avro so I can >start >interacting with some of my Avro-based Python services. > >Is there any document I should be looking at other than the specification >on >avro.apache.org and maybe the Python code to start implementing? > > >Cheers > >- R. Tyler Croy >-------------------------------------- > Code: http://github.com/rtyler > Chatter: http://twitter.com/agentdero > http://identi.ca/dero
