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 32243 - AbstractFileSystem is doing too expensive refreshes which are unnecessary
Summary: AbstractFileSystem is doing too expensive refreshes which are unnecessary
Status: VERIFIED FIXED
Alias: None
Product: platform
Classification: Unclassified
Component: Filesystems (show other bugs)
Version: 3.x
Hardware: All All
: P1 blocker (vote)
Assignee: rmatous
URL:
Keywords: PERFORMANCE
Depends on:
Blocks: 32065
  Show dependency tree
 
Reported: 2003-03-24 16:10 UTC by Petr Nejedly
Modified: 2008-12-22 21:38 UTC (History)
2 users (show)

See Also:
Issue Type: DEFECT
Exception Reporter:

Attachments
release35 diff (2.84 KB, patch)
2003-03-27 13:07 UTC, rmatous 		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 Petr Nejedly 2003-03-24 16:10:54 UTC 
For events where the FS knows what happened is
refresh too expensive, e.g. during on createData()
Comment 1 Petr Nejedly 2003-03-24 16:12:06 UTC 
It is more visible on VCS filesystems as their "refresh" is more
expensive.
It strongly affects e.g. internal compilation.
Comment 2 rmatous 2003-03-26 10:52:55 UTC 
Fixed in trunk.
/cvs/openide/src/org/openide/filesystems/AbstractFileObject.java,v 
<--  AbstractFileObject.java
new revision: 1.86; previous revision: 1.85

/cvs/openide/src/org/openide/filesystems/AbstractFolder.java,v  <-- 
AbstractFolder.java
new revision: 1.66; previous revision: 1.65
Comment 3 Petr Nejedly 2003-03-27 10:22:51 UTC 
OK, internal compiler is much faster now e.g. on JavaCVS
Comment 4 Petr Nejedly 2003-03-27 11:02:13 UTC 
We should put this into 3.5, it greatly improves usefullnes of
internal compiler.
Radku, will you ask for review/integration?
I've looked at the changes and they seems OK to me.
Comment 5 _ ttran 2003-03-27 12:35:22 UTC 
because 3.5 is focused on performance and this one seems to be a
significant performance issue, I agree we should try to get the fix
in. -> P1

Petr N or someone, please provide some numbers so we can see how the
users are affected
Comment 6 rmatous 2003-03-27 13:07:09 UTC 
Created attachment 9562 [details]
release35 diff
Comment 7 Petr Nejedly 2003-03-27 14:36:04 UTC 
Unqualified numbers are ~120s vs. ~60s for compiling openide/src
on JavaCVS FS. Exact numbers are hard to take as the variance
of the result is quite big (remember, it does a lot of file access).
Comment 8 rmatous 2003-03-28 08:21:13 UTC 
Fix: Avoid to call refresh after new child (FileObject) was created.
And directly add new child in parent's map.
Comment 9 Petr Nejedly 2003-03-31 08:47:43 UTC 
The patch is OK
Comment 10 _ ttran 2003-04-01 08:15:03 UTC 
approved for 3.5 by release coordinator
Comment 11 rmatous 2003-04-01 08:23:03 UTC 
Fixed in release35.