Use http-1.0.0 for http-cache-tower #70
Conversation
Adds cache mode fixes, overridden_cache_mode method, and impl Default for CacheMode
|
Pushed a new branch that I think might ease this implementation a bit, at least I think it might in regards to running the async code blocks https://github.com/06chaynes/http-cache/tree/remove-async-trait |
|
One prerequisite to merging this will be the Reqwest update to use hyper v1.0.0, which thankfully seems to be underway here https://github.com/seanmonstar/reqwest/tree/hyper-v1 After updating in my local branch everything but the http-cache-reqwest crate seems to be good to go, so once Reqwest itself is updated it should be fairly straightforward to complete on this end. |
| match this.state.as_mut().project() { | ||
| ResponseFutureProj::Init => { | ||
| let request_parts = this.req.clone().into_parts().0; | ||
| let fut = this.cache.before_request(&request_parts); |
There was a problem hiding this comment.
@06chaynes, I took this line from the commented-out code in your branch, but there's no before_request method on cache. Assuming it's because the previous branch was out of date, what should be called here instead?
If you have an async version of the code, I can probably translate it to the hand-rolled future
There was a problem hiding this comment.
@SebRollen yeah in that branch I was experimenting with a different implementation that I decided (at least for now) not to go with, though I did bring over a handful of changes.
Here is the implementation for Reqwest that has the logic as it currently is https://github.com/06chaynes/http-cache/blob/develop/http-cache-reqwest/src/lib.rs#L212
First the Middleware trait implementation is constructed, then the request is checked to determine if it's cacheable. If it isnt cacheable we just run the run_no_cache method of the HttpCache struct, handle the response normally, update the two cache headers, and pass it through. If it is cacheable we run the run method of the HttpCache struct and pass through the result.
There was a problem hiding this comment.
pushed a commit to update the core and other crates http and http-cache-semantics versions
|
Sweet, reqwest has been updated. Just need a new reqwest-middleware release and this particular change should be about ready |
|
@06chaynes It seems that |
That comment was in regards to updating http, which has already been merged in |
WIP branch for updating the
refactorbranch to usehttp-1.0.0, per our discussion in #4.If accepting the
Clonebound onReqBodyis acceptable, it simplifies the passing of the request to the future quite a lot.If this looks promising to you, I can also try to tackle updating the future into a state-machine so that we can poll other futures inside of it. I know the
refactorbranch is quite far behind themainbranch though, so if it makes more sense to work off of that one, let me know. I might need some pointers in terms of which methods need to be called inside the future since I'm not too familiar withhttp-cacheinternals.