>>>>> On Tue, 25 Apr 2017 17:07:21 -0400, "Martin Mulholland" <mmulholl_at_us.ibm.com> said:
MM> Have not seem anything on this since Stuart raised the issue. Have there
MM> been any additional (private) conversations on this topic?
In the interest of striking a balance between timeliness and diplomacy,
we did pre-advise licensees shortly before sharing this with the entire
EG.
MM> Like Stuart, we consider this ALPN solution as a hack.
Noted.
MM> But, if we consider Stuart's suggestion of not requiring secure
MM> http/2 in Servlet 4, would that mean to make it optional?
We will not be making HTTP/2 optional in Servlet 4.0.
MM> Or, would we consider removing it altogether from Servlet 4?
We will not be removing HTTP/2 from Servlet 4.0.
MM> Any of these approaches would affect the TCK/CTS testing of Servlet
MM> 4.
The only change in the plan for TCK/CTS testing for Servlet 4.0 is the
decision to only test h2c from the TCK/CTS. This precipitated the
change in the Servlet spec to require implementations to support h2 and
h2c.
MM> Any possible change in direction here would definitely affect our
MM> plans going forward and we'd like to close on this issue.
From Oracle's perspective, the issue of what to do about ALPN in Servlet
4.0 is closed as follows:
* On the servlet container implementation side, we are using this approach
>>>>> On Tue, 18 Apr 2017 09:08:53 +1000, Stuart Douglas <sdouglas_at_redhat.com> said:
SD> 1) Override the boot class path (Jetty ALPN, and the proposed
SD> Oracle solution)
We require Servlet 4.0 implementations to support h2 and h2c.
* On the TCK client side we are using the approach of only testing on h2c.
I hope this removes any doubt about the resolution.
Thanks,
Ed
--
| edward.burns_at_oracle.com | office: +1 407 458 0017