Summary: | hovering allocates megabytes in seconds | ||||||
---|---|---|---|---|---|---|---|
Product: | [Eclipse Project] JDT | Reporter: | Adam Kiezun <akiezun> | ||||
Component: | Text | Assignee: | JDT-Text-Inbox <jdt-text-inbox> | ||||
Status: | RESOLVED WONTFIX | QA Contact: | |||||
Severity: | normal | ||||||
Priority: | P3 | CC: | philippe_mulet | ||||
Version: | 2.1 | Keywords: | performance | ||||
Target Milestone: | --- | ||||||
Hardware: | PC | ||||||
OS: | Windows 2000 | ||||||
Whiteboard: | |||||||
Attachments: |
|
Description
Adam Kiezun
2002-11-28 10:02:47 EST
Created attachment 2565 [details]
118 megabytes allocated by 5 seconds of hovering
i meant 18, not 118 97% caused by codeSelect Again, this is a consequence of some eager activation from JDT/UI side. please clarify what you mean by 'eagerly activate' when you hover, we call codeSelect - which allocates all this memory what can we do differently? CodeSelect isn't a free operation, if UI activates it too often, when there is no actual expectation from a user (i.e. while cursor is still moving) then it is pointless. moving to text for comment and further action. can codeSelect be called less often on hovering? The call of codeSelect is result of the SWT mouse hover event. Currently, there is not way to change its timing. The computation of the hover info, i.e. the call of codeSelect, is performed in a background thread. If the mouse moves on while this background thread is active, it would be useful to cancel the thread. Most of the time is spent in codeSelect. Thus, having codeSelect cancelable would be useful because it would ensure that at most one computation is running at any time. Haven seen any reaction from core. |