As I've noted previously, our Aeson instances should really implement the newer toEncoding(http://hackage.haskell.org/package/aeson-1.4.7.1/docs/Data-Aeson.html#v:toEncoding) interface, which is much faster than encoding via toJSON. We should be able to derive these generically with genericToEncoding in much the same way we've been using genericToJSON.
I also noticed as I was tracking down the test failures on the REST side of the library that there seem to be some hand-coded Aeson instances there that can be replaced with generically derived instances, much as we just did with the hand-coded instances I originally wrote for the WebSocket types.
These changes should be fairly straightforward. I'll try to get to them soon.
As I've noted previously, our Aeson instances should really implement the newer
toEncoding(http://hackage.haskell.org/package/aeson-1.4.7.1/docs/Data-Aeson.html#v:toEncoding) interface, which is much faster than encoding viatoJSON. We should be able to derive these generically withgenericToEncodingin much the same way we've been usinggenericToJSON.I also noticed as I was tracking down the test failures on the REST side of the library that there seem to be some hand-coded Aeson instances there that can be replaced with generically derived instances, much as we just did with the hand-coded instances I originally wrote for the WebSocket types.
These changes should be fairly straightforward. I'll try to get to them soon.