RFR [12]: java.net.http.HttpClient hangs on 204 reply without Content-length 0

[Michael McMahon]

Hi Daniel,

On 15/10/2018, 12:53, Daniel Fuchs wrote:
> Hi Michael,
>
> On 15/10/2018 11:54, Michael McMahon wrote:
>> Could I get the following fix reviewed please.
>>
>> http://cr.openjdk.java.net/~michaelm/8211437/webrev.1/index.html
>
> Looks good in general.
>
> In MultiExchange.java:
>
> 236             T nullBody = cf.get();
>
> Though technically the body should be available by the
> time we reach this line, since you completed the subscriber
> just before, we can't really make any assumption on the
> implementation of the subscriber.
>
> So for the sake of robustness we should probably use
> getBody().handle(...) to complete `result`
> rather than calling cf.get();
>

Yes, that would be better.

>
> Also I wonder what should happen if a body is present:
>
> Should we simply read it instead?
> Because if we don't then we should close the connection (HTTP/1.1)
> or reset the stream (HTTP/2) - which probably means getting back
> to the concrete  exchange implementation to make sure that
> happens.
>

I see that as a protocol error. So, rather than attempting to read the
body, I think we should ensure that the request fails (which it does
if a content length or transfer encoding field is present). The connection
also needs to be closed (or stream reset) which I need to check is being 
done.

Thanks,
Michael

> best regards,
>
> -- daniel
>
>>
>> Thanks,
>>
>> Michael.
>