DefaultHasher is not documented as being consistent. It's best to not
trust that the resulting hash value is consistent even across different
executions of the same binary and even more so across different
versions.
These appear to only be used by examples in the old/ directory which has
not compiled for a long time. Not sure why the contents of that
directory are still around.
No more need to macro_use serde. Order the imports consistently (albeit
somewhat arbitrary), starting with items from this crate, followed by
std, followed by external crates.
The current code uses standard blocking i/o operations (std::fs::*) this
is problematic as it would block the entire futures executor waiting for
i/o.
This change is a major refactoring to make the token storage mechansim
async i/o friendly. The first major decision was to abandon the GetToken
trait. The trait is only implemented internally and there was no
mechanism for users to provide their own, but async fn's are not
currently supported in trait impls so keeping the trait would have
required Boxing futures. This probably would have been fine, but seemed
unnecessary. Instead of a trait the storage mechanism is just an enum
with a choice between Memory and Disk storage.
The DiskStorage works primarily as it did before, rewriting the entire
contents of the file on every set() invocation. The only difference is
that we now defer the actual writing to a separate task so that it does
not block the return of the Token to the user. If disk i/o is too slow
to keep up with the rate of incoming writes it will push back and
will eventually block the return of tokens, this is to prevent a buildup
of in-flight requests. One major drawback to this approach is that any
errors that happen on write are simply logged and no delegate function
is invoked on error because the delegate no longer has the ability to
say to sleep, retry, etc.
This upgrade Hyper to v0.12 and updats to code to work for it. It has
being done with the minimum code change and so the logic is still
aukward for the futures model. This should be addressed in later commits
but I did not want to compilcate an already large commit.
Recently, commits were made to remove support for using rustls/ring
in lieu of OpenSSL to avoid having to add those as dependencies in
environments where OpenSSL already exists.
However, yup-oauth2 is being used in some environments where
*OpenSSL* doesn't exist. So, retaining the option to build
it with an openssl-free stack is key.
Note, the change to hyper-native-tls is fine and actually
unrelated, because this is often happening on Windows or
Mac enviornments, where the stack that will link to is
*not* necessarily OpenSSL either.
For example, the particular breakage that prompted this
CR is a build of yup-oauth2 on iOS. In that environment,
hyper-native-tls uses Apple's Security Framework,
but OpenSSL libraries are not actually present on all
devices/simulator environments. So a link against
a mix of Security Framework and rustls make sense,
there.
Also, update to a newer version of rustls/ring while we're at it.
This removes the need for the remaining C interfaces.
Building any rust library with openssl adds a bunch
of depends such as foreign types via ffi and pkgconfg.
You are also required to have a prebuilt openssl.
Cross building and keeping up to date should be easier
with a pure rust implementation.
This bumps rustls to 0.6.1 which bumps webpki-roots.
We need this because the current version of webpki 0.8.0
has been removed from crates.io.
I created the 0.6.x rustls for hyper 0.10 support.
In future we should upade to hyper 0.11 which is rusls 0.8.0+
The current Git master doesn't build with Rust 1.17 because the
version of Rustls used on master doesn't built with Rust 1.17.
* Update the base64 dependency to ensure the version with a buffer
overflow fix is used.
* Update hyper-rustls to one that uses a version of Rustls that builds
with Rust 1.17, and that uses *ring* 0.9+, to ensure there are no
problems with accidentally linking multiple versions of *ring*, to make
the build easier to set up (especially on Windows), and to keep things
working once *ring* 0.7.* gets yanked.
* Update the rust-openssl dependency while we're at it.
With Rust 1.15, proc macros have been stabilized. Therefore
custom build scripts are not required anymore.
This commit removes all the previous machinery and the
need for nightly.
