the-sett / elm-aws-codegen Goto Github PK

Was asked about the iot service on slack. That's a REST-JSON one.

There is an issue with refined strings not matching their regexes correctly. This could probably be solved by escaping the regex correctly. However, after discussing it, decided it would be better to not do this field-by-field validation of requests and responses. It makes the API awkward to use, and is unusual for a client stub to do this anyway, since the AWS services themselves validate their inputs.

Remove the refined Int and String types, and replace with plain Ints and Strings.

To achieve this:

1. Add options to the Elm.Lang code generation module in Salix. It should be an L3 processor.
2. Use the same options in the Elm.Json.Coding module.
3. Make these options common to the above processors.

By doing the above, the use of refined types becomes a parameter of the code generator - can put it back at a later date if desired. Or possible validate responses, but not requests, etc.

Based on the cognito service file the DateType is a timestamp which is internally represented as int (Unix secons) for AWSCognito, specifcally the expiration field of identity credentials

As per the the sdk code here, the timestamp can be iso string , date object or a unix timestamp

Some of the AWS specs redefine basic types, so the code ends up with things like:

type alias Username = String 

Or even:

type alias String = String

Should not create aliases for basic types. Remove these aliases and just use the basic type directly.