> Now I'm thinking that if response content is going to be in effect a
> generator, then request content should be in effect an accumulator. 
> Indeed, they can be *actually* SRFI 158 generators and accumulators. 
> Time to make a Bitbucket page for this: 
> <https://bitbucket.org/cowan/r7rs-wg1-infra/src/default/HttpRequest.md>.

Thanks for writing up a concrete proposal. It looks reasonable.

I'm concerned we don't understand HTTP APIs and HTTP 2 well enough. I
don't think we have any people on this list who have made non-trivial
use of asynchronous HTTP for real (I hope I'm wrong). The present
confusion with keyword arguments or lack thereof around SRFI and
R7RS-large in general adds to the mix.