Explicitly removing the idled out thread from the stack,
rather than calling tryExecute().
Side benefit is that we are now removing idled out threads
that are least recently used.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
* Issue #4550 XmlConfiguration argument matching
Improve argument matching by:
+ rejecting obviously non matches (with allowance for unboxing)
+ sorting methods so that derived arguments are tried before more generic (eg String before Object)
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4550 XmlConfiguration argument matching
Improve argument matching by:
+ can unbox from any Number to any Number
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4550
Do not check the assignability of the arguments. Instead rely on the order of the methods.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4550
unbox test no longer required
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4550
Simplified test
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4550
Cleanup comparator
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4550
Cleanup comparator
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Fixes#4577 IPAccessHandler in context
Fixes and tests #4577 IPAccessHandler in context by using target instead of pathInfo for path matching.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Tests #4577 IPAccessHandler target
Updates from review.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4577 IpAccessHandler NPE
Match on full URI path rather than target.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
Changed the way the test directory was created: it was based on a
millisecond timestamp, but the tests run fast and it was possible that
two tests were creating the directory within the same millisecond.
Signed-off-by: Simone Bordet <simone.bordet@gmail.com>
Fixes#4575 Stopping Reserved Thread by removing the `isRunning` check
from `reservedWait`. The main run loop is also simplified to improve
`isRunning` checks before the thread is put on the stack. Javadoc
improved to explain each step.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
Fixed#4550 named parameters with a moderate refactor.
The named parameter matching was duplicated, only considering number of args and not applied to call arguments. This refactor puts all the behaviour in common methods and reorders the arguments to match parameters.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
Fix#4541 by initially allocated a header buffer of `min(_config.getResponseHeaderSize(), _config.getOutputBufferSize())`
Only allocate a buffer of `getResponseHeaderSize` if an overflow results.
This should have no effect on the majority of responses where `getOutputBufferSize` is greater than `getResponseHeaderSize` other than the cost of a min operation.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
Added option for #4529 to control showing the servlet in an error page.
Error configuration really needs a module, but will add in 10 with a refactor.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
* Issue #4533 Hard close from Dispatcher
Signed-off-by: Greg Wilkins <gregw@webtide.com>
#4533 Do hard close from Dispatcher so response wrappers may intercept close.
* Issue #4533 Hard close from Dispatcher
Signed-off-by: Greg Wilkins <gregw@webtide.com>
#4533 improve test after review
* Issue #4533 Hard close from Dispatcher
Some renaming of methods to make it clear that softClose should only be used as part of sendError handling. If softClose is used by other components, then sendError can be prevented from setting the error status.
Signed-off-by: Greg Wilkins <gregw@webtide.com>
Simone Bordet
Hirotaka Ikoma
Joakim Erdfelt
Greg Wilkins
Lachlan
Olivier Lamy
Sergey Tselovalnikov
Ludovic Orban
Jan Bartel
Jesse McConnell