http-client <-> canonical request

olorin commented 8 years ago

On top of #4. (diff)

Convert between http-client Requests and CRequests per the spec draft. A few things:

The conversion can fail in the http-client -> CRequest direction for various reasons.
Certain restrictions are placed on the Request (e.g., must use one of the methods explicitly specified in RFC 2616). This isn't because I think it's unsafe to sign requests which don't meet these standards, it's just that I don't see any reason this would be useful and so I'm erring on the side of being restrictive for now. A bunch of these will probably open up once we've got the basic protocol in place and can better see the consequences of some of the decisions here.

olorin commented 8 years ago

@markhibberd @erikd-ambiata @thumphries @nhibberd

olorin commented 8 years ago

Ignore failing jenkins builds, they're post-boris but before I actually deleted the jobs.

thumphries commented 8 years ago

This all looks fairly reasonable to me 🉐

erikd-ambiata commented 8 years ago

:ocean: I'm ok with this.

haskell-mafia / zodiac