Monadic Retry Policies about req HOT 5 CLOSED

I'm not sure how common this need is, yet it'd cause a lot of breaking changes. If we need to allow access to MonadHttp instance then HttpConfig should be parametrized over some m to contain RetryPolicyM m inside...

from req.

Yeah, I'm not entirely convinced that it's worth it versus simply "disabling" the built-in retry policy (httpConfigRetryPolicy = limitRetries 0 or something) and then using the retry package directly with my own RetryPolicyM (parameterized with my application's type) and retrying.

It wouldn't be much extra work, just that I'd have to replicate some of the req functionality on my own.

from req.

How about changing from

httpConfigRetryPolicy :: RetryPolicy

to

httpConfigRetryPolicy :: RetryPolicyM IO

since Req is IO anyways (and MonadHttp requires MonadIO). This retry policy can then be lifted internally to Req via natTransformRetryPolicy liftIO.

This wouldn't require any breaking changes, but it would fix the largest drawback (imo) of the current approach: fullJitterBackoff cannot be used due to its requirement of IO.

from req.

@ramirez7 Do you want to prepare a PR?

from req.

👍 yep I'm on it

from req.