On Wed, 12 Jan 2022 16:59:14 GMT, Julia Boes <jb...@openjdk.org> wrote:
> This new test confirms that the jwebserver does not wait indefinitely for a
> request to arrive, but instead closes the connection when the maximum request
> time is reached. To facilitate this, Exchange::run is amended to process
> cancelled keys in the case where the current thread is the dispatcher thread
> (which is the case for the jwebserver).
>
> Testing: tier 1-3 all clear.
src/jdk.httpserver/share/classes/sun/net/httpserver/ServerImpl.java line 525:
> 523: logger.log(Level.TRACE, "exchange started");
> 524:
> 525: if (dispatcherThread != null && dispatcherThread ==
> Thread.currentThread()) {
Hello Julia,
Minor thing - With the `dispatcherThread == Thread.currentThread()` check, the
`dispatcherThread != null` would be redundant and perhaps can be removed?
test/jdk/com/sun/net/httpserver/simpleserver/jwebserver/MaxRequestTimeTest.java
line 104:
> 102: sendHTTPRequest(); // server expected to respond successfully
> 103:
> 104: p.destroy();
Since the `sendXXXRequest()` methods above could potentially throw exceptions,
do you think it would be better to destroy the process in a finally block? Or
maybe manage the process start and destroy as testng before/after lifecycle
methods?
-------------
PR: https://git.openjdk.java.net/jdk/pull/7052
