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 19938 - Deadlock during switching workspaces
Summary: Deadlock during switching workspaces
Status: CLOSED FIXED
Alias: None
Product: platform
Classification: Unclassified
Component: Window System (show other bugs)
Version: 3.x
Hardware: PC Linux
: P2 blocker (vote)
Assignee: David Simonek
URL:
Keywords: THREAD
Depends on:
Blocks:
 
Reported: 2002-01-30 12:40 UTC by Milan Kubec
Modified: 2008-12-23 14:32 UTC (History)
4 users (show)

See Also:
Issue Type: DEFECT
Exception Reporter:

Attachments
full thread dump (13.80 KB, text/plain)
2002-01-30 12:41 UTC, Milan Kubec 		Details
Should not this be fixed in ExplorerManager - the code used to be there, but was commented out for some reason (639 bytes, patch)
2002-02-08 09:53 UTC, Jaroslav Tulach 		Details | Diff
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Milan Kubec 2002-01-30 12:40:36 UTC 
I got deadlock during switching WS. I exited IDE in Debugging WS and then
started again. Deadlock occured after switching from Debugging WS to another
(Editing). The deadlock occured when editor window came up. I can provide
userdir, because it seems reproducible.
Comment 1 Milan Kubec 2002-01-30 12:41:11 UTC 
Created attachment 4492 [details]
full thread dump
Comment 2 Jan Zajicek 2002-01-31 16:44:34 UTC 
CCing Yarda to propose, how to fix this deadlock. Should
deserialization run in separate thread or the read data should be
instantiated later?

It isn't reproducible, however it can happen under normally
circumstances. Fix will not be trivial. From these reasons this will
not go into 3.3.1.
Comment 3 mslama 2002-02-01 13:41:59 UTC 
FolderChildren calls waitFinished() on default Request Processor. Jarda,
could FolderChildren use its private request processor?
Comment 4 Jaroslav Tulach 2002-02-08 09:53:30 UTC 
Created attachment 4617 [details]
Should not this be fixed in ExplorerManager - the code used to be there, but was commented out for some reason
Comment 5 friv9821 2002-02-14 10:35:08 UTC 
The deadlock also occurs when executing, if execution is set to swith 
to workspace running. I've not been able reproduce it, but it happens 
quite frequently, several times every day. Since I changed execution-
setting NOT to swith, it has never happened.

(Running nb 3.3.1 on winnt sp 6)
Comment 6 David Simonek 2002-02-20 16:24:00 UTC 
Jesse, as cvs annotate shows, you commented out posting of
initialization into request processor. Now it seems that we have
troubles with it (look at deadloceked stack traces)

Can you explain why you did it and if I can revert it? 

1.19 (jglick   25-Apr-00):
Comment 7 Jesse Glick 2002-02-20 21:34:39 UTC 
1.19 did not do it, that was reindenting for open source - check the
log. The code in question was added in 1.11 by Yarda :-) and the
RequestProcessor line was commented out from the start, it seems.
Comment 8 Jaroslav Tulach 2002-02-21 14:30:24 UTC 
Looks I was playing with RP and then decided to invoke the code
directly. Probably ok to move to RP...
Comment 9 David Simonek 2002-02-21 15:52:14 UTC 
ok, I'll try it :-)
Comment 10 David Simonek 2002-02-21 17:01:24 UTC 
committed into main trunk.
Comment 11 Jan Chalupa 2002-04-09 16:20:51 UTC 
Risky fix. Will not be fixed for FFJ 4.0 (Orion).
Comment 12 Jan Chalupa 2002-05-02 10:31:25 UTC 
Waiver approved.
Comment 13 Milan Kubec 2002-10-17 10:34:59 UTC 
Cannot reproduce.
Comment 14 Quality Engineering 2003-07-01 16:22:20 UTC 
Resolved for 3.4.x or earlier, no new info since then -> closing.
Comment 15 Quality Engineering 2008-12-23 14:32:52 UTC 
This issue had *1 votes* before move to platform component