Tracking issue for libz blitz evaluation of reqwest

[dtolnay]

This is the tracking issue for the evaluation performed by the libs team this week.

• Expose RedirectAttempt and RedirectAction types
• Hide trait impls that are inaccessible to users
• Send and Sync in return type of Error::get_ref
• Switch to &mut self builder
• Move try_ macro to error.rs
• Use realistic error handling in example code
• Include an "Errors" section in documentation of functions that can return error
• Decide how to handle JSON serialization errors
• Comment why unwrap() won't fail
• Include html_root_url attribute
• Response doc improvements
• Examples for the Body constructors
• Error doc improvements
• Example for RequestBuilder::headers
• Examples for RequestBuilder::body
• Conform to the recommended pattern for sealed traits
• Reimplement StatusCode using associated constants
• Reimplement Method using associated constants
• Stabilize some of the foundational types in a separate crate

[theduke]

I don't know if this was discussed in the meeting,
but what are your thoughts regarding hyper async and stabilizing request?

Should 1.0 wait for a stable release of hyper async?

Or should there already be a SyncClient with a backwards compatible API once hyper async arrives?

[seanmonstar]

It was mentioned that since an async client will need to expose Future, and the futures crate won't stabilize before reqwest, then reqwest 1.0 will just have a sync client, and and reqwest_async will live seperately as an unstable thing. When it can stabilize, it could be merged into reqwest and released as v2.0.

[seanmonstar]

That said, to reduce breakage of reqest v1 -> v2, the Client naming issue is relevant here.

[little-dude]

Lots of these issues are doc or examples. I'm thinking about writing examples using https://httpbin.org. Would that be good, or did you have other ideas in mind?

[seanmonstar]

It's fine for the example to refer to httpbin, just because the example isn't in a fn main() {}, so we don't have the doctests actually making requests.

[budziq]

Hi large part of docs (including part of landing page) are missing links when referencing symbols. Should these be corrected (assuming yes from log crate evaluation)?

[brson]

#1196 was superceded by the existing hyperium/hyper#894 issue, which still needs to be fixed before reqwest 1.0