Just for the record, I've started a driver for OkHttp [1]. It currently works and makes all tests (mock and live) pass, including the PATCH ones.
The driver integration tests, are failing randomly (:() and still haven't found a pattern. Tests only fail sometimes (not always the same tests) and still haven't been able to successfully debug them. If someone could give a hand and help fixing the test, we could be able to try the driver with other providers and see if it actually works as expected "outside the Marconi api". Ignasi [1] https://github.com/nacx/jclouds/tree/okhttp/drivers/okhttp On 13 December 2013 18:43, Andrew Phillips <[email protected]> wrote: >> Given this, do you think we should revert the patch [2]? It only >> "extends" the reflection stuff to properly set the method in HTTPS >> connections, so in practice I think the code could remain. If we keep >> the patch, at least we would be able to use PATCH requests with a body >> over HTTPS, as the sun.net.www.protocol.http.HttpURLConnection only >> fails then the protocol is HTTP. >> >> WDYT? > > > Fine with keeping it in. It doesn't break anything (as far as we know), and > reverting will still leave us with "reflection code without a purpose". > > Looks like we have some serious engineering ahead of us for 1.8.0. Do we > know anything about the support for different HTTP methods in the Java 7 or > 8 libraries..? > > ap
