This patch does a bunch of refactoring, mostly around error types, but it also
paves the way to allow `Codec` to be used standalone.
* `Codec` (and `FramedRead` / `FramedWrite`) is broken out into a codec module.
* An h2-codec crate is created that re-exports the frame and codec modules.
* New error types are introduced in the internals:
* `RecvError` represents errors caused by trying to receive a frame.
* `SendError` represents errors caused by trying to send a frame.
* `UserError` is an enum of potential errors caused by invalid usage
by the user of the lib.
* `ProtoError` is either a `Reason` or an `io::Error`. However it doesn't
specify connection or stream level.
* `h2::Error` is an opaque error type and is the only error type exposed
by the public API (used to be `ConnectionError`).
There are misc code changes to enable this as well. The biggest is a new "sink"
API for `Codec`. It provides buffer which queues up a frame followed by flush
which writes everything that is queued. This departs from the `Sink` trait in
order to provide more accurate error values. For example, buffer can never fail
(but it will panic if `poll_ready` is not called first).
Header fields and data frames can potentially contain sensitive data.
This change omits these from Debug output which reduces the chance that
this ends up in logs.
Malformed requests and responses should immediately result in a
RST_STREAM. To support this, received header frames are validated and
converted to Request / Response values immediately on receipt and before
buffering.
Use a temporary private fork of tokio-rustls that uses Rustls 0.12
until tokio-rustls 0.4 is released.
This upgrades, among other things, *ring* to 0.12, which will ensure
that it still builds in the Rust 1.20 release coming this week even if
backward-compatibility-breaking changes to rustc aren't fixed before
the release.
Because, you might think that each linked list has exclusive access to
the next pointer, but then there is an edge case that proves otherwise.
Also, debugging this kind of thing is annoying.
Restructure send flow control such that sending data is always accepted by `Stream`. Data frames will be buffered until there is available window to send them. Producers can monitor the available window capacity to decide if data should be produced.
With this change, h2 can build and run without any manual configuration steps for -msvc targets. Previously manual installation of OpenSSL libraries was required.
tokio-io 0.1.3 has been published so the "replace" for tokio-io 0.1.2 is no longer useful and it breaks the build with some (but apparently not all) versions of Cargo.
Update the dependency to tokio-io 0.1.3 since that is the version that's what's been verified to work.
Client::poll_ready ensures that the connection is ale to to initiate a new request stream to the remote server. When the server is at capacity, a task is stored to be notified when capacity is available.
