This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 167260 - [67cat] [ruby] Code Completion takes excessive time to show windows
Summary: [67cat] [ruby] Code Completion takes excessive time to show windows
Status: CLOSED DUPLICATE of bug 167193
Alias: None
Product: editor
Classification: Unclassified
Component: CSL (API & infrastructure) (show other bugs)
Version: 6.x
Hardware: PC Linux
: P2 blocker (vote)
Assignee: mslama
URL:
Keywords: PERFORMANCE
Depends on:
Blocks:
 
Reported: 2009-06-17 19:36 UTC by esmithbss
Modified: 2009-06-30 08:14 UTC (History)
6 users (show)

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
Profile Trace. Started just prior to code completion invokation. Ended just after display of rdoc dialog. (57.67 KB, application/octet-stream)
2009-06-17 19:37 UTC, esmithbss
Details

Note You need to log in before you can comment on or make changes to this bug.
Description esmithbss 2009-06-17 19:36:13 UTC
[ BUILD # : 6.7rc3 ]
[ JDK VERSION : 1.6.* ]

I just ran some experiments and have profiled the user invoked
code-completion dialogs showing up.  For a pre-filtered display (some
characters typed before invoking code-completion), it took 20+
seconds to display the dialog.

According to the profile trace, fully 18417ms of this is in the
Parsing & Indexing Loop. (see attached profiler trace).
Comment 1 esmithbss 2009-06-17 19:37:28 UTC
Created attachment 83721 [details]
Profile Trace.  Started just prior to code completion invokation.  Ended just after display of rdoc dialog.
Comment 2 Erno Mononen 2009-06-17 19:54:06 UTC
Raising to P2, that is unacceptably slow.
Comment 3 Erno Mononen 2009-06-17 20:54:48 UTC
For one reason or other, the index appears to be *really* slow on your machine - each query seems to take almost a 
second to complete. Is this with a clean userdir? If not, does running with a clean userdir make any difference? I'll 
investigate more tomorrow, but it does look like the problem is in the indexing infrastructure.
Comment 4 Erno Mononen 2009-06-18 09:20:21 UTC
Reassigning to csl after discussing this with Vita, seems to be a recent regression caused by the tl indexer blocking 
the index.
Comment 5 Vitezslav Stejskal 2009-06-18 10:44:01 UTC
This is the same problem as issue #167193. I believe we have fix suitable for 6.7.

*** This issue has been marked as a duplicate of 167193 ***
Comment 6 Marian Mirilovic 2009-06-30 08:14:54 UTC
v/c