Automated Deployment stages
Currently the process of managing different versions of lambda and interaction models is very painful. I bet most teams end up just having multiple instances of the same skill to simulate common deployment stages like "int" "test" "beta" or "production".
I'd suggest a feature which automatically maps different stages of the release process to different versions of the lambda and language model. In order to do this, to following would be required:
- Support versioning and aliasing of the Alexa models (etc) similar to AWS functionality
- Support automated mapping between those alias, i.e. a skill tagged as "alias-1" will call it's lambdas with "alias-1"
- Allow testers to test against specific versions/aliases of a skill
- Allow developers to select which alias will be used for the "beta" and "production" stages of the skill
On a side note, this might also introduce away to help devs with the release process. A skill alias could be submitted to certification instead, allowing developers to switch the production version between any previously certified aliases or versions.