Jörg Kubitz

If we find out why there is any benefit to have any delay that would be needed. Otherwise we can just make UI more responsive.

> We've already found out that the delay is here to prevent useless computation from happening too early in case of repeated fast interactions. did we? any proof?

Multiple developers making the same assumption is no solid support. There have been so many useless waits in eclipse that have been successfully totally removed that you should at least...

Avoiding concurrency at all is a well known pattern to avoid concurrency issues. And if (to be tested) it even helps to improve other things is simply two slain with...

normaly such updates should be throttled by Throttler see https://github.com/eclipse-platform/eclipse.platform.ui/commit/56fd3f80f14125f357779a3e96c2c8eaa9facf93#diff-e2ef7af122667f19ba26e15fc1cda3712bdaf6d8efbd72123b944772b067d7d7R144 please check if just keeping the throttletime to 500ms solves your issue.

For clarification: I want to use 0 initial delay, but still a positive waiting period between events is ok. The Throttler already provides a distinction of the both concept. And...

the delay as used in org.eclipse.jface.text.contentassist.AdditionalInfoController controlles that the completion proposal YELLOW page pops up 500ms later ( i always thought its so expensive to calculate). Is there any benefit...

> You mean the code element information? the right pop up with additional information, containing the javaDoc in this example.

``` java.lang.AssertionError: No new shell found expected: but was: at org.junit.Assert.fail(Assert.java:89) at org.junit.Assert.failNotEquals(Assert.java:835) at org.junit.Assert.assertEquals(Assert.java:647) at org.eclipse.ui.genericeditor.tests.CompletionTest.findNewShell(CompletionTest.java:220) at org.eclipse.ui.genericeditor.tests.CompletionTest.testMoveCaretBackUsesAllProcessors_bug522255(CompletionTest.java:251) ``` ``` Wrong job order: arrays first differed at element [0];...

> In most cases, computing extra information for a completion item is expensive Please see the analysis in linked issue: the "timeout" did not prevent computing that information anyway. Please...