On 03/06/2019 22:29, Rémy Maucherat wrote:
> How about this as an idea:
>
> --- a/java/org/apache/coyote/AbstractProtocol.java
> +++ b/java/org/apache/coyote/AbstractProtocol.java
> @@ -905,6 +905,10 @@
> }
> }
> }
> + // The handler will initiate any further I/O
> + if (wrapper.hasAsyncIO()) {
> + state = SocketState.LONG;
> + }
> }
> } while ( state == SocketState.UPGRADING);
>
>
> Essentially, it is saying if the upgrade handler is async, it will take
> care of triggering any further reads that may be necessary.
>
> Initial test results are promising.
>
> Worth trying.
Close, but it wasn't quite right. It is the UpgradeHandler that needs to
be tested.
This works with Linux. Just running the tests on Windows...
http://people.apache.org/~markt/patches/2019-06-03-h2-v1.patch
Mark
PS Having had to get my head around the non-blocking changes - kudos.
Nice work.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
