I used to think that aws-clj-sdk was a bit of NIH, but the approach seems more right to me now. Amazonica gets quite far, but there is feeling of a stronger grip in going to the APIs directly.

Can aws-clj-sdk provide better docs and arglists than amazonica?

I really disliked that the arguments in amazonica are like string-1 string-2 input-stream-1… difficult to say what they are

although I understand the technical reason (can’t get the argnames by reflecting the client Java class)

1/ shapes-as-specs would show up in doc output (doc for free) 2/ I believe that @viesti is working on exploiting these fine json files https://github.com/aws/aws-sdk-ruby/blob/master/apis/lambda/2014-11-11/docs-2.json

oh yes, I think I did some line-breaking stuff for the docs

now I remember

:thinking_face: are the AWS APIs implemented in Java, names like "UploadFunctionRequest$FunctionZip" seem like inner class names to me

Maybe but here the $ seems to denote more a field than an inner class (FunctionZip is a Blob)

docs are extracted, and for example cider spec browsing works quite nicely

might be even more descriptive than python/boto kwargs version

we just need to support more carrier protocols, to have complete success :)