RFR: 8223714: HTTPSetAuthenticatorTest could be made more resilient

[Jaikiran Pai]

On 30/08/19 5:40 PM, Jaikiran Pai wrote:
> Can I please get a review and a sponsor for a patch for
> https://bugs.openjdk.java.net/browse/JDK-8223714?
>
> The patch is available as a webrev at
> http://cr.openjdk.java.net/~jpai/webrev/8223714/1/webrev/
>
> This issue and the patch relates solely to a test infrastructure class -
> HTTPTestServer and touches no main source code. The patch involves the
> following:
>
> - A straightforward fix in the readLine method which prevents the
> java.lang.StringIndexOutOfBoundsException reported in that JIRA.
>
> - Additional changes to the run() method of the server to make the
> connection handling and processing of this (test) server a bit more robust.
>
>     - The run method now accepts a connection and if while processing
> the client request on that connection, runs into a IOException, then
> closes the client connection and then continues to accept any subsequent
> connections (this wasn't the case previously, where the server would
> just abort and no longer accept any more connections).
>
>     - Any non IOException(s) are logged and the server goes on to accept
> any subsequent requests.
>
>     - Whether or not to accept any more connections is now handled by
> the "stop" flag which gets set when the server stop is requested.
>
>     - Any failure to "accept()" a connection stops the server from
> accepting any more requests. This is also now clearly logged as a log
> message.
>
>     - The bulk of the parsing and processing of the client request has
> been moved out of the run() method into a new private
> processRequestAndWaitToComplete method, for better code clarity.
>
> I haven't been able to reproduce the original exception reported in the
> JBS issue, before doing any of these changes. However, I have run the
> existing HTTPSetAuthenticatorTest,

... and the HTTPTest and they have passed.

-Jaikiran